The disputation — a special type of cooperative argumentative dialogue
Modality in Learner German - A corpus-based study investigating expressions of modality in...
Transcript of Modality in Learner German - A corpus-based study investigating expressions of modality in...
Modality in Learner German
A corpus-based study investigating modal expressions in
argumentative texts by British learners of German
A thesis submitted for the degree of Doctor of Philosophy
by
Ursula Maden-Weinberger
(1. Staatsexamen, M.A.)
Department of Linguistics and English Language
Lancaster University
June 2009
Abstract
This thesis examines, by way of a corpus-driven investigation, usage patterns of
modal expressions by British learners of German as a foreign language. Modality
plays a pivotal role in all texts and discourses that deal with the discussion or
examination of practical or theoretical problems and is crucial for the process of
argumentation and reasoning. It is therefore an integral part of foreign language
learners‟ writing at university level. The study is more comprehensive than previous
approaches, as it incorporates the full semantic range of modal meanings as well as
all the different types of modal expressions in the German language, i.e. modal verbs,
modal adverbials and modal lexical verbs, modal infinitives and the subjunctive
mood. The investigation is methodologically based on a multiple-comparison
approach involving a corpus of learner German (CLEG), which has been compiled for
the study, and several other L1 and L2 native speaker corpora. The results indicate
that learners exhibit in their argumentative writing specific patterns of modality use
that are different from those of native speakers and at various levels of proficiency.
On the whole, learners tend to overuse modal expressions. The specific overuse and
avoidance patterns of epistemic modal verbs are strongly linked to L1/L2 form
meaning equivalents (e.g. MUST – MÜSSEN). Where modal verbs are avoided,
modal adverbials are overused to fill the semantic gap. Modal adverbials tend to
develop into fossilised formulaic sequences, where the expression of modality is
extrapolated from the rest of the utterance in a sentence-initial matrix clause, usually
of the “es ist...” („it is...‟) type. The development of the use of the subjunctive mood is
indicative of the learners‟ growing proficiency; however, learners even at the
advanced stage exhibit difficulties in the morphological distinction between indicative
and subjunctive verb forms. On top of these findings on learner modality, the results
from this study can corroborate previous research on general characteristics of
learner language, specifically the tendency to be more explicit than native speakers,
to overuse personalised expressions that indicate subjective writer involvement, and
to exhibit a less formal, more speech-like register in the learners‟ written
argumentative productions.
Declaration
This thesis is my own work and has not been submitted in substantially the same
form for the award of a higher degree elsewhere.
Some results displayed in chapter 5 have been published in the following article:
Maden-Weinberger, U. (2008). “Modality as Indicator of L2-Proficiency? A corpus-
based investigation into advanced German interlanguage”. Walter, M. and Grommes,
P. (eds). Fortgeschrittene Lernervarietäten: Korpuslinguistik und
Zweitspracherwerbsforschung / Advanced Learner Varieties. Corpus Linguistics and
Research into Second Language Acquisition. Berlin: Niemeyer
Ursula Maden-Weinberger
23 June 2009
1
TABLE OF CONTENTS
1. INTRODUCTION 10
1.1 Focus of the study 11
1.2 Innovations, research questions, aims and claims 13
2. THEORETICAL CONSIDERATIONS ON MODALITY 17
2.1 The concept of modality 17
2.1.1 Definitions of modality 18 2.1.2 Epistemic and deontic modality 19 2.1.3 Other types of modality and other means of expressing modality 22 2.1.4 Terminological differences 24
2.2 Modality in the German language 27
2.2.1 Requirements of a theoretical approach for the investigation into learner modality 27
2.2.2 Brinkmann’s account of the modal system 29
2.3 Linguistic means for expressing modality 33
2.3.1 Modal verbs 33 2.3.1.1 Formal characteristics 33 2.3.1.2 Semantic characteristics 38
2.3.1.2.1 MÜSSEN 39 2.3.1.2.2 KÖNNEN 40 2.3.1.2.3 WOLLEN 41 2.3.1.2.4 SOLLEN 42 2.3.1.2.5 DÜRFEN 43 2.3.1.2.6 MÖGEN 43 2.3.1.2.7 WERDEN 45
2.3.2 Modal adverbials 47 2.3.2.1 Brinkmann’s (1971) account of modal adverbs 48 2.3.2.2 Zifonun et al.’s (1997) account of modal adverbials 49
2.3.3 Modal lexical verbs and quasi-modal verbs 54 2.3.4 Modal infinitives HABEN ZU and SEIN ZU 56 2.3.5 Subjunctive mood (Konjunktiv) 62
2.3.5.1 The basic functions of Konjunktiv I and Konjunktiv II 62 2.3.5.2 Syncretism between subjunctives and indicative 65 2.3.5.3 Usage contexts for Konjunktiv I and Konjunktiv II 69
3. MODALITY IN SECOND LANGUAGE ACQUISITION RESEARCH 75
3.1 Modality in naturalistic acquisition 75
3.2 Modality in instructed acquisition 79
2
4. METHODOLOGY 87
4.1 The corpus linguistic basis of the study 87
4.1.1 Learner corpus research 89 4.1.2 Contrastive Interlanguage Analysis and the Multiple Comparison
Approach 91
4.2 Investigating German learner modality within the multiple comparison approach 94
4.2.1 The German learner corpus: CLEG 94 4.2.2 The comparable native speaker corpus: KEDS 98 4.2.3 The general native speaker corpus: LIMAS 100 4.2.4 The translation corpus for contrastive analysis: INTERSECT 100
4.3 Analysing modality in the learner corpus 103
4.3.1 Selecting and retrieving modal expressions 104 4.3.2 Semantic categorisation and annotation 106 4.3.3 Analytical techniques 116
4.3.3.1 Quantitative and qualitative investigations 117 4.3.3.2 Interpretation of results 117 4.3.3.3 Statistics 119
5. RESULTS 122
5.1 Modal verbs 122
5.1.1 Epistemic modal verbs 127 5.1.1.1 Overview 127 5.1.1.2 Modal verbs without English form/meaning equivalents: Evidential
WOLLEN and SOLLEN, DÜRFTE, MÜSSTE 130 5.1.1.3 Modal verbs with English form/meaning equivalents: MÜSSEN,
KÖNNEN/KÖNNTE, WERDEN, MÖGEN, SOLLTE 131 5.1.1.4 MÜSSEN (compelling conclusion) 133 5.1.1.5 KÖNNEN and KÖNNTE (assumption of possibility) 134 5.1.1.6 WERDEN (confident assumption) 136 5.1.1.7 MÖGEN (assumption of possibility) 143 5.1.1.8 SOLLTE (tentative conclusion) 144
5.1.2 Non-epistemic modal verbs 145 5.1.2.1 Overview 145 5.1.2.2 MÜSSEN (duty/compulsion) and MÜSSTE (suggestion) 148 5.1.2.3 SOLLEN (instruction/obligation) and SOLLTE (suggestion/advice) 153 5.1.2.4 KÖNNEN and KÖNNTE (possibility and ability) 158 5.1.2.5 DÜRFEN and KÖNNEN (permission) 161 5.1.2.6 WOLLEN (volition) 164 5.1.2.7 MÖCHTE (attenuated volition) and MÖGEN (affection/liking) 165
5.2 Periphrastic modal expressions 168
5.2.1 Overview 168 5.2.2 Epistemic periphrastic modal expressions 173
5.2.2.1 Assertive periphrastic modal expressions 173 5.2.2.2 Modally qualifying periphrastic expressions 185
3
5.2.2.3 Negative periphrastic modal expressions 197 5.2.3 Non-epistemic periphrastic modal expressions 200
5.3 Modal infinitives 213
5.3.1 SEIN ZU 214 5.3.2 HABEN ZU 217
5.4 Subjunctive mood (Konjunktiv) 218
5.4.1 Overview 218 5.4.2 Lexical distribution of Konjunktiv II 222 5.4.3 Context types of Konjunktiv II 224 5.4.4 Konjunktiv II in conditional clauses 226
6. SUMMARY AND DISCUSSION 234
6.1 Summary of results 234
6.2 Discussion 237
6.2.1 Overuse of modal expressions 237 6.2.2 Ranking of modal verbs 240 6.2.3 Form/meaning equivalents in learners’ native and target language 241 6.2.4 Correspondences between usage patterns of modal verbs and
periphrastic modal expressions 247 6.2.5 Difficulties with morphological and semantic distinction of indicative and
Konjunktiv II modal verb forms 249 6.2.6 Personalised expressions 254 6.2.7 Range of modal expressions and formulaic sequences 256 6.2.8 Konjunktiv II as indicator of learner proficiency 263
6.3 Concluding remarks 266
7. CONCLUSION 268
BIBLIOGRAPHY 275
APPENDIX 286
Appendix A 286
Appendix B 290
Appendix C 292
Appendix D 297
4
LIST OF FIGURES
Figure 1: Position of a corpus study of modality in learner German within the wider research context ..................................................................................................... 13
Figure 2: Contrastive Interlanguage Analysis (CIA) framework (Granger 2002:12) ................................................................................................................................. 91
Figure 3: Multiple-Comparison Model for learner corpus research (adapted from Tono 2002a) ............................................................................................................... 92
Figure 4: Multiple-comparison model for the study of modality in learner German .................................................................................................................................... 94
Figure 5: Overall frequencies of modal verbs .......................................................................... 124
Figure 6: Individual modal verb frequencies ........................................................................... 125
Figure 7: Epistemic modal verb frequencies ........................................................................... 129
Figure 8: Non-epistemic modal verb frequencies .................................................................. 147
Figure 9: Types of grammatical subject/passive constructions with non-epistemic SOLLTE ............................................................................................................ 156
Figure 10: Frequencies of indeterminate KÖNNEN (annotation category “U”) ........... 160
Figure 11: Comparison of modal verbs and periphrastic modal expressions (PMEs) .................................................................................................................................. 169
Figure 12: Overall frequencies of epistemic and non-epistemic modal expressions ......................................................................................................................... 170
Figure 13: Epistemic vs. non-epistemic PMEs ........................................................................... 172
Figure 14: Frequencies of assertive PMEs ................................................................................... 173
Figure 15: Frequencies of modally qualifying periphrastic expressions ........................ 185
Figure 16: Frequencies of PME categories “subjective assumption” versus “tentative conclusion” .................................................................................................. 188
Figure 17: Frequencies of PMEs affecting negation................................................................. 198
Figure 18: Frequencies of non-epistemic PMEs ........................................................................ 201
Figure 19: Frequencies of Konjunktiv II context types ........................................................... 224
Figure 20: Distribution of Konjunktiv II across context types as percentages of overall Konjunktiv II frequencies ............................................................................... 225
Figure 21: Conditional clause patterns as % of overall frequencies of conditionals ........................................................................................................................ 226
5
LIST OF TABLES
Table 1: Aims and claims of the presented study 15
Table 2: Overview of terminology for epistemic and non-epistemic modality 26
Table 3: Present tense indicative and Konjunktiv I 66
Table 4: Present tense indicative and Konjunktiv I 66
Table 5: Preterite indicative and Konjunktiv II (weak verbs) 66
Table 6: Preterite indicative and Konjunktiv II (strong verbs) 67
Table 7: Sein/haben/werden in indicative, Konjunktiv I and Konjunkti II 67
Table 8: Size and Subcorpora of the Corpus of Learner German (CLEG) 97
Table 9: The INTERSECT translation corpus 101
Table 10: Tagset for the annotation of modal expressions 108
Table 11: Tagset for the annotation of modal expressions (translated) 110
Table 12: Overall frequencies of modal verbs in learner and native speaker corpora 123
Table 13: Log-likelihood values for differences in modal verb frequencies between CLEG, KEDS and LIMAS 123
Table 14: Statistical significance of differences in modal verb frequencies between learner year groups and KEDS 126
Table 15: English equivalents to German modal verbs (epistemic) 128
Table 16: Mutual correspondence values (MC) for German and English epistemic modal verb equivalents 132
Table 17: Epistemic MÜSSEN 133
Table 18: Error rate for epistemic KÖNNTE/KONNTE 135
Table 19: WERDEN with present-time reference 136
Table 20: WERDEN used in “habitual” and “general events” contexts 137
Table 21: WERDEN with future-time reference 138
Table 22: Epistemic MÖGEN in concessive contexts 143
Table 23: English equivalents to German modal verbs (non-epistemic) 146
Table 24: Mutual correspondence values (MC) for German and English non-epistemic modal verb equivalents 147
Table 25: Non-epistemic indicative MÜSSEN (MUSSTE) and Konjunktiv II MÜSSTE 149
Table 26: Non-epistemic NICHT MÜSSEN as prohibition 153
Table 27: Errors non-epistemic SOLLEN as suggestion 154
Table 28: Error-corrected frequencies of non-epistemic SOLLEN and SOLLTE 155
Table 29: Non-epistemic SOLLEN in Questions 156
Table 30: Percentage distribution of indicative and Konjunktiv II forms of KÖNNEN 159
6
Table 31: Error rate for non-epistemic KÖNNTE/KONNTE 160
Table 32: DÜRFEN and KÖNNEN as permission 162
Table 33: Negated DÜRFEN 163
Table 34: Negated DÜRFEN including *NICHT MÜSSEN 163
Table 35: MÖCHTE in 1st person signposting utterances 166
Table 36: Overall frequencies of periphrastic modal expressions in learner and native speaker corpora 169
Table 37: Lexical diversity of periphrastic modal expressions 171
Table 38: Raw frequencies (freq. per 10,000 words in brackets) of purely assertive PMEs 174
Table 39: Raw frequencies (freq. per 10,000 words in brackets) of positive evaluative-assertive PMEs 175
Table 40: Raw frequencies (freq. per 10,000 words in brackets) of reservedly positive evaluative-assertive PMEs 176
Table 41: Raw frequencies (freq. per 10,000 words in brackets) of negative evaluative-assertive PMEs 177
Table 42: Sentence-initial leider 177
Table 43: Evaluative-assertive PMEs along other dimensions 178
Table 44: Raw frequencies (freq. per 10,000 words in brackets) of evidence-focused assertive PMEs 180
Table 45: Colligational sequence …Tatsache, dass… 181
Table 46: Top five evidence-focused assertive PMEs 183
Table 47: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting a compelling conclusion 186
Table 48: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting subjective assumptions 189
Table 49: Colligational sequences ich denke, dass…/ich glaube, dass.../ich bin der Meinung, dass… 190
Table 50: Sentence initial ich denke/ich glaube/meine Meinung 190
Table 51: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting tentative conclusions 192
Table 52: Colligational Sequence es scheint 193
Table 53: Colligational Sequence mir scheint/es scheint mir 193
Table 54: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting assumptions of possibility 195
Table 55: Distribution of adjective möglich and noun Möglichkeit 196
Table 56: Colligational sequences es ist möglich/es gibt die Möglichkeit 196
Table 57: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting evidential meaning 197
Table 58: Raw frequencies (freq. per 10,000 words in brackets) of PMEs achieving a hedging of negations 198
7
Table 59: Raw frequencies (freq. per 10,000 words in brackets) of PMEs achieving an intensification of negations 199
Table 60: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting duty or compulsion 202
Table 61: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting instructions/obligations 203
Table 62: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting suggestions/advice 203
Table 63: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting volition 204
Table 64: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting attenuated volition/wishes 204
Table 65: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting general affection or liking 206
Table 66: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting possibility due to circumstances 207
Table 67: Types of –bar-adjectives in the PME category “possibility due to circumstances” 208
Table 68: Colligational sequence es ist (un)möglich… 209
Table 69: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting possibility due to personal abilities 211
Table 70: Raw frequencies (freq. per 10,000 words in brackets) of PMEs denoting permission 212
Table 71: Overall frequencies of modal infinitive constructins in learner and native speaker corpora 214
Table 72: Raw figures of SEIN ZU across semantic meaning categories 214
Table 73: Colligational sequences es ist (ADJ) zu V, dass… 215
Table 74: Raw frequencies of infinitive verb types in SEIN ZU constructions 216
Table 75: Raw frequencies of colligational sequence sein adj zu inf 216
Table 76: Overall frequencies of Konjunktiv I forms in learner and native speaker corpora 219
Table 77: Overall frequencies of Konjunktiv II forms in learner and native speaker corpora 219
Table 78: Error frequencies: Konjunktiv instead of indicative verb forms 220
Table 79: Distribution of Konjunktiv II forms across verb categories 222
Table 80: Konjunktiv II verb categories as percentages of overall Konjunktiv II frequencies 222
Table 81: Konjunktiv II Lexical verb types and raw frequecy figures 223
Table 82: Overall frequencies of conditionals 226
Table 83: Error rate for conditional clause pattern P(I) – A(KII) 230
Table 84: Error rate for conditional clause pattern P(KII) – A(KII) 230
8
Table 85: Errors in conditional clause pattern P(KII) – A(KII) 232
APPENDIX
Appendix A: Examples of learner texts
Appendix B: Periphrastic modal expressions
Appendix C: Data tables corresponding to overview charts
Appendix D: Examples of handouts with “Redemittel” (useful phrases) and exercises in
preparation for writing argumentative texts
9
Acknowledgements
Academically, my thanks go to my supervisor, Dr. Andrew Wilson, for his help and
support throughout the process of this PhD and to Marjorie Wood for sorting out all
the administrative issues. I would also like to thank the Economics and Social
Research Council for providing funding for this PhD. A special thanks goes to the
students who gave their permission to use their material in my research and to the
colleagues in the Department of European Language and Cultures, especially Birgit
Smith, Jane Wilkinson, Manuela Piller, Ulrike Tancke and Judith Menzel, who helped
collect the data and also supported me through their friendship.
Personally, I would like to thank my parents for their never ending love and support
whatever I do. Thanks also to Sue, Ross and Alex – especially for all the extra
babysitting in the last year. I thank all my friends in Lancaster for inspiring
conversations, motivating pep-talks and pure and simple distraction. Special thanks
go to Prof. Maurice Kirby and Jo Hardman for letting me squat in their office.
Finally, and most importantly, I am deeply indebted to Will, whose love and care
always pull me through, and to Corin, who always makes sure that I know that work is
never that important. Last but not least, thank you Annika for putting a very definitive
deadline on this PhD.
10
1. Introduction
This thesis examines, by way of a corpus investigation, usage patterns of modal
expressions by British learners of German as a foreign language. Corpus linguistics
is a set of tools and methods for the analysis of computerised databases of written or
spoken texts (corpora) which supports linguistic research across all branches of the
subject. With the relatively recent upsurge of corpus linguistic methods, it has
nowadays also been widely recognised that corpora of learner language provide
insightful information that can further our understanding of second language
acquisition through a detailed, yet quantitatively grounded, analysis of learners‟
language production (cf. e.g. Lorenz 1999; Granger et al. 2002; Myles 2005; Walter
and Grommes 2008). The seminal work by Sylivane Granger and her colleagues on
the ICLE corpus (International Corpus of Learner English) has certainly established
the comparative investigation of over-, under- and misuse patterns in different groups
of English language learners and native speakers as an accepted methodology for
studying learner corpora (Granger 1996). This so called „contrastive interlanguage
analysis‟ has become an important field of study for researchers interested in learner
language, but also for teachers and materials designers who strive to incorporate the
knowledge about what kind of language learners produce in their decisions about
what to teach and how to teach it.
However, while a wide variety of general and specialised1 corpora is available for
English, resources for the German language are far more restricted. Large, publically
accessible corpora of native speaker German are available, for example, from the
Institut für Deutsche Sprache in Mannheim (Institute for the German Language) or
the DWDS (Deutsches Wörterbuch der deutschen Sprache des 20. Jahrhunderts –
„Digital Dictionary of the 20th Century German Language‟), but they do not, in any
respect, reflect the breadth and diversity of English resources (for an overview of
German corpora cf. Lemnitzer and Zinsmeister 2006; Geyken 2009). For German
learner language, the situation is even more limited. At the time this study was
undertaken, no corpus of learner German of reasonable size existed that fulfilled the
strict design and sampling criteria that are crucial in order to obtain reliable
1 „General corpora‟ are large-scale corpora that attempt to reflect a specific language or variety in all its contexts of use. They consist of a large number of written and/or spoken texts to represent a cross-section of text-types, genres, registers and subject fields and are often used as the „norm‟ against which particular aspects of language use are measured. „Specialised corpora‟ are usually smaller corpora where the content is sampled from a particular language variety, subject area, genre or group of language users, often designed with a specific research purpose in mind.
11
information2. The first major achievement of this study is therefore the compilation of
a German learner corpus (CLEG) that is comparable in design and size to the
subcorpora of the ICLE corpus, which is the most widely used corpus for studies of
learner language (cf. e.g. Granger 1998a). Secondly, it was necessary to compile a
comparable native speaker corpus (KEDS) in order to carry out contrastive
interlanguage analyses, as this type of corpus had equally not existed previously for
German. This corpus matches CLEG very closely with regard to text type and age
and expertise of writers. These new resources allow for a multitude of comparisons
between learners at different stages of proficiency and native speakers and open up
novel ways of gathering information about German as a foreign language.
The fact that the corpus captures a language other than English brings the added
advantage of allowing for comparisons across languages in order to investigate
universal patterns of learner language. The ICLE is made up of subcorpora of learner
English by learners with different L1 backgrounds. It is therefore possible to
determine certain features of learner writing as L1 dependent or independent. With a
corpus of learner German, comparisons with English learner language can also be
made that enable researchers to discover and establish certain features of learner
language as independent of a particular L2. This adds a new dimension of evidence
to learner language studies, which has the potential for giving wider implications in
second language acquisition research.
1.1 FOCUS OF THE STUDY
One of the major advantages of a corpus is its universality. Once it is compiled, the
corpus can serve as the empirical basis for any number of investigations into different
aspects of language. Within the scope of this thesis, of course, only one of these
aspects can be focused upon. As CLEG consists of argumentative texts written by
student learners, the central topic investigated here is modality.
Modality, as such, is an intriguing concept as it centres around the effect of a
single fundamental human trait, namely to be able to conceive of things as being
otherwise. Modality is grounded in the fact that human beings can, and often do, think
and behave as though things might be other than they actually are and “such a
worldview appears to constitute an essential part of the fabric of our everyday lives”
2 This situation has fundamentally changed with the release of the Falko corpus (Fehlerannotiertes Lernerkorpus – error annotated learner corpus) (cf. Lüdeling et al. 2005 and section 4.1.1 for details). This corpus was, however, not available at the time of data gathering and analysis for the present study.
12
(Perkins 1983:6). Modality therefore plays a pivotal role in all texts and discourses
that deal with the discussion or examination of practical or theoretical problems and is
crucial for the process of argumentation and reasoning (Zifonun et al. 1997:1911;
Redder 2001:314).
The language learners that provide the empirical basis for the present study are
students of German as a foreign language in the British university system. In this
context of formal language instruction, the learners regularly perform tasks that
require reasoning and the discussion of problems. As the learners produce
argumentative texts3 as part of their university assessment, mastering expressions of
modality thus forms an integral and important part of the students‟ writing ability,
especially in the more advanced proficiency groups. Furthermore, the various
linguistic manifestations of modality are high frequency phenomena in the German
language (e.g. all modal verbs are ranked in the 100 most frequent words). Learners
are therefore confronted with them early on in the acquisition process and encounter
them frequently as soon as they begin to deal with authentic texts. However, many
learners appear to have difficulties in this area, even at an advanced stage, not just in
terms of grammatical and semantic errors, but also in terms of the rhetoric and
stylistic impact of certain modal expressions.
The present study therefore aims at clarifying, with the help of a learner corpus,
how learners of German make use of modal expressions, where the difficulties lie
and what the underlying reasons for particular usage patterns could be. This implies
that the study can be positioned within current research from three angles that
constitute the foundation for this study in terms of theory as well as research
methodology: modality, second language acquisition (SLA) and corpus linguistics.
First of all, the study has to be based on a theoretical understanding of the concept of
modality including the linguistic manifestations of the German modal system.
Secondly, as it is a study of learner language, it has to be placed within the context of
SLA interlanguage studies and finally, it will be argued that corpus analysis provides
a viable and fruitful methodological framework for investigating modality features in
learner language. The diagram below illustrates that, while there is a high interest
and research activity in all three of these areas (darker shades of grey indicate higher
research output), the present study is the first to combine all of them together:
3 The notion of an argumentative text-type is based on the framework of Werlich (1976), who distinguishes five text-types: narration, exposition, description, argumentation and instruction; according to Biber (1988) these are determined by text internal features, rather than by genre.
13
FIGURE 1: POSITION OF A CORPUS STUDY OF MODALITY IN LEARNER GERMAN WITHIN THE WIDER
RESEARCH CONTEXT
1.2 INNOVATIONS, RESEARCH QUESTIONS, AIMS AND CLAIMS
Apart from the general gap in the research, which has been demonstrated in Figure 1
and which is filled by the present study, there are two specific innovations:
i) The study is based on a corpus of learner German. While a variety of learner
corpora are nowadays available for English, a German learner corpus of the scale
and design that was compiled for this study did not previously exist.
ii) The study includes an exhaustive list of modal expressions, which is essential for
a comprehensive study of modality in learner language. Previous research studies
either restricted their scope to one category of modality (usually epistemic) or to
one means (usually modal verbs). The current study is innovative insofar as it
includes all categories as well as all means of modality.
Corpus Study of Modality in
Learner German
SLA
Studies
Modality
Studies
Corpus
Studies
14
Within this research design, the following research questions are addressed:
1. Which modal expressions do learners use to express which modal meanings at
different stages of proficiency?
2. Where and how do these patterns of use differ from native speaker use of modal
expressions, i.e. do learners overuse or avoid certain modal expressions? Which
areas of modality are prone to misuse/errors?
3. What reasons can be indentified for the differences in use (L1 transfer,
intralingual difficulties, teaching materials)?
These research questions are addressed within an empirical research study that
sets out to pursue the following aims and seeks to substantiate the following claims:
Aims Claims
1. Identify a suitable theoretical
framework for the study of learner
modality. This framework has to be
based on semantic categorisations
and include all means of modal
expression.
A comprehensive study of learner
modality cannot be restricted to only one
category of modality or one type of modal
expressions. The analysis of previous
research will show that all modal
expressions have to be taken into
account to give an insightful picture
2. Identify a suitable methodological
framework for the study of learner
modality, that allows for the analysis of
a wide variety of modal expressions
The multiple-comparison approach to
learner corpus data provides meaningful
information for the study of learner
language. A corpus-driven approach
allows for the detection of indicative
features of learner language
3. Identify patterns of modality use in
learner language
Learners differ in their use of modal
expressions from native speakers but
also from learners at different stages of
proficiency
15
4. Identify possible reasons for the
deviations in learner use of modal
expression on the basis of previous
modality research and current SLA
theories
The differences in learners‟ usage of
modality can be explained by and provide
corroborating evidence for a number of
universal and L1/L2 specific features of
learner language.
TABLE 1: AIMS AND CLAIMS OF THE PRESENTED STUDY
The thesis is structured in the following way:
Chapter 2 introduces the theoretical concept of modality. This starts off with a
general overview of the different notions and meanings that have been associated
with the term „modality‟ (2.1), before reviewing different accounts of modality
specifically in the German language (2.2). On the basis of this review a choice is
made for the account that forms the basis of the empirical investigation. This account
is then presented in detail (2.3).
Chapter 3 gives an overview of previous research on modality in second language
acquisition. Reference will be made to influential work on the development of modal
expressions in both natural and instructed language acquisition.
In Chapter 4, the methodological basis for the study is explained and justified. This
includes a short rationale of corpus research and learner corpus research in general
(5.1.1 and 5.1.2 respectively) and then details the specific methodological framework
that is adopted for the present investigation: the multiple-comparison approach
(5.1.3). In the second part of the methodology, this framework is applied to the study
of German learner modality and all necessary data gathering and analysis
procedures are introduced (5.2).
Chapter 5 constitutes the presentation of results. These are grouped, first of all, by
type of modal expression (i.e. modal verb, periphrastic modal expression, modal
infinitives, Konjunktiv mood). After brief overviews of these groups, detailed results
are given for individual modal expressions. Links to results for other modal
expressions are made where possible.
In Chapter 6, these results are discussed and evaluated in the context of previous
research on learner modality and current theories of second language acquisition.
Chapter 7 concludes the thesis with a brief review of the aims and claims set out
at the beginning and some remarks on further research.
16
In summary, the design of this study of modality in argumentative learner German
seeks to provide a new and insightful picture into a specific aspect of learner
language that has previously not been investigated with such comprehensiveness. It
may also yield information which points to more universal, i.e. language-independent,
features of learner language that could be useful for the development of improved
teaching methods and materials. The study of modality in learner German is therefore
deemed to be a valuable enterprise not just for the advancement of knowledge of this
particular aspect but also in the wider context of second language acquisition
research.
Before starting with the theoretical considerations on modality, I would like to give
some explanations on notation as used in this thesis: modal verbs and modal
infinitives are given in block capitals (e.g. WOLLEN, HABEN ZU) where they refer to
the lemma rather than specific verb forms. A distinction is, however, made between
modal verbs in the indicative and the subjunctive (e.g. KÖNNEN (ind.) vs. KÖNNTE
(subj.)), as these distinctions are crucial to the present investigation. German
examples and terminology are in italics, as are English examples in the running text.
Double quotes are used for citations, single quotes are used for terminology, i.e.
concepts and categories. Translations are given for all German terminology and
examples in order to make clear for English-speaking readers the points that are
illustrated in the German examples. In many cases simple one-to-one translations
(e.g. of the modal verbs or one-word modal adverbials) are sufficient. Some
translations, however, include glosses where the English translation does not show
the lexical or grammatical issue under discussion. I have tried to stay as closely to
the German text as possible, e.g. modal verbs are translated as modal verbs in all
cases where this is possible. There are, of course, always alternatives with different
modal expressions, but as the translations here are intended purely for the benefit of
non-German readers, this option is the most useful for the intended illustrative
purpose.
17
2. Theoretical Considerations on Modality
2.1 THE CONCEPT OF MODALITY
The concept of modality has come a long way as a topic of interest to formal logic -
and nowadays computer science - as well as philosophy and linguistic reasoning. But
despite the fact that the study of modality originated with Aristotle in the 4th century
B.C. and continued to produce a wealth of publications throughout antiquity, the
Middle Ages and modern times, it was only in the 20th century that linguists started to
concentrate their efforts on the study of modality and the functions of modal features
in individual languages and from a language typological point of view.
However, as Dittmar and Reich (1993:IX) claim, much of these earlier efforts
constituted “theoretical outlines”, and “new insights [...] have been gained by
introspection” while “empirical research on concepts and verification of hypotheses
have [...] been widely neglected” [my emphasis].
Even though the latter part of this quote is no longer quite justified in the face of a
series of empirical studies on modality (some of which will be discussed in this
section), the majority of the literature on modality is indeed more concerned with
theoretical attempts to define modality as a linguistic category. On this basis attempts
are made at defining the range of linguistic phenomena to be included under the term
„modality‟ in any given language.
This has proved immensely difficult and is probably the reason why many of the
accounts of modality concentrate on theoretical considerations: As Öhlschläger
(1984:229) points out in his research review of modality in German4: “While the
expressions commonly subsumed under the term „modal‟ can certainly be considered
one of the most interesting phenomena, it is equally certain to be one of the most
difficult to describe not only for the German language but also for other languages.”5
To this date there is no generally agreed linguistic definition of modality, which also
results in considerable disagreement over linguistic expressions of modal meanings.
In the following sections, the basic issues and controversies concerning the concept
of modality will be summarised briefly. (For an extensive discussion see e.g. Palmer‟s
(1986) “Mood and Modality”, which constitutes a comprehensive attempt at a
4 For ease of reading, quotes from German literature are translated into English in the running text; the original quotes are given in footnotes.
5 "Die gemeinhin als modal bezeichneten Ausdrücke gehören sicherlich zu den interessantesten, aber
ebenso sicher auch zu den am schwierigsten zu beschreibenden Erscheinungen des Deutschen wie auch anderer Sprachen“.
18
systematic and principled cross-linguistic description of modality.) On the basis of this
outline, which will deal mainly but not exclusively with approaches to modality in the
German language, I will explain which definition will be adopted as the theoretical
foundation for the present empirical study and, consequently, which linguistic items
will be included in the investigation.
2.1.1 Definitions of modality
As the plural in this heading suggests, there is a plethora of definitions of modality
that reflect not only the various fields of research that are concerned with the topic,
such as formal logic, philosophy and linguistics, but also the different schools of
thought within these fields. However, since they all use the same term it is assumed
that there is some common notion of what „modality‟ means, at least on a very basic
and broad level.
As mentioned above, a lot of the groundwork on modality originates in formal logic,
where modal logic is concerned with the logic of possibility, necessity and
contingency. In a pioneering work on modality within this logical framework, von
Wright (1951:1-2) postulates four „modes‟:
alethic (modes of truth)
epistemic (modes of knowing)
deontic (modes of obligation)
existential (modes of existence)
which are linked through the two “central notions of traditional modal logic” (Lyons
1977:787) – possibility and necessity.
This kind of categorisation, however, is set up in order to resolve issues in formal
logic. It does not necessarily follow from this that these categories can simply be
transferred to describe the systems of modal expressions and relations in a linguistic
context. Palmer (1986:11) argues, for example, that in language “there is no formal
grammatical distinction […] between alethic and epistemic modality”, i.e. the ways to
“state what is logically true and what the speaker believes, as a matter of fact, to be
true”. It is therefore neither possible nor necessary to keep this distinction in a
linguistic account of modality. It has, however, become apparent for linguistic models
of modality that it is useful to make a fundamental distinction between two types of
modality: epistemic and deontic.
19
2.1.2 Epistemic and deontic modality
The distinction between epistemic and deontic modality is based on the fact that most
modal verbs in many languages (including English and German, but also non-Indo-
European languages) can have both epistemic and deontic interpretations. Without
context – which usually serves to resolve ambiguity – the following sentence can
theoretically be read in terms of “knowledge/belief/opinion of the speaker” (epistemic)
(cf. Lyons 1977:793) or, equally, in terms of “permission/obligation of the subject”
(deontic) (ibid:823). It works the same way in English and German6:
(1)
He must understand that we
mean business.
Er muss verstehen, dass wir es ernst meinen.
epistemic „Surely he understands that we mean business‟
deontic „It is essential that he understand that we mean business‟
Within this framework, which has clearly been developed out of the logical models,
the relationship between the two types of modality – epistemic and deontic – depends
firstly on the link between possibility and necessity and secondly on an assumed link
between possibility and permission on the one hand and necessity and obligation on
the other (cf. Steele et al. 1981). The first proposition (the link between possibility and
necessity) is usually demonstrated through negation: If we take the proposition Er ist
krank (He is ill), the expression of an epistemic necessity of that proposition (2) is
equivalent to the negative possibility of the inverse proposition (2a) or inverse to the
negative possibility of the same proposition (2b):
(2) Er muss krank sein. 'He must be ill.' necessity that p: „It is necessarily the case that he is ill.‟
(2a) Er kann nicht gesund sein.
'He cannot be healthy.' negative possibility of negative p: „It is not possible that he is not ill = that he is healthy‟
(2b) Er kann nicht krank sein. ‟He cannot be ill.‟ negative possibility that p: „It is not possible that he is ill‟
6 Example from Coates (1983:16).
20
The second proposition about the relationships between possibility and permission
on the one hand and necessity and obligation on the other is often assumed without
justification or explanation and seems to be borne out of the fact that in many
languages the same modal verbs can be used for epistemic and deontic modality –
which actually renders the explanation circular.
However, the question does arise as to what the semantic connection between
these two types of modality is to produce such an overlap in forms of modal verbs7. A
compelling explanation is advocated by Sweetser (1982, 1990) within a cognitive
framework. She argues on the basis of first language acquisition research and the
historical development of modal verbs that “the epistemic world is understood in
terms of the sociophysical world” (1982:492) by cognitive extension. This means that
deontic modality refers to various forces or barriers in the socio-physical world, while
epistemic modality, through metaphorical processes, refers to parallel forces in the
world of reasoning (premises, previous knowledge, assumptions). Therefore,
example (1) (He must understand that we mean business.) above in its deontic sense
means that the subject is obliged by the speaker‟s or some other authority to carry
out the proposed action. By extension, the epistemic reading means that the
premises in the speaker‟s mind (through assumption or previous knowledge) oblige
them to draw the conclusion that the proposed action is being carried out. This
cognitive approach is also shared by Heine, Claudi and Hünnemeyer (1991:176):
“The relation between deontic and epistemic modality can in fact be described as
being metaphorical in nature, involving a transfer between two domains of
conceptualization.”
In a similar vein, Brünner and Redder (1983:46) propose that, methodically, modal
verbs are to be treated as an integrative system, in which what they call „inferential‟
(epistemic) and „non-inferential‟ (non-epistemic) meanings can be determined within
an action-theoretical framework. This means that the meanings of modal verbs are
systematically related to elements of action processes, which allows for the fact that
modal verbs (unlike other lexical expressions) only acquire their concrete meaning ad
hoc in a given discourse context (ibid:39). In that way, epistemic uses of modal verbs
7 There is extensive discussion whether this overlap in forms is a manifestation of monosemy or polysemy. Monosemantic approaches (e.g. Ehrmann 1966, Kratzer 1978, Brünner and Redder 1983, Brinkmann 1971) assert a „basic meaning‟ for each modal. Supporters of polysemantic approaches (e.g. Leech 1971, Calbert 1975) on the other hand deal with modals in terms of separate categories, which are assumed to be discrete. To date, “neither approach is wholly satisfactory” (Coates 1983:9) and Coates argues that “it is not simply a case of adopting or rejecting discrete categorisation, or of preferring a monosemantic or a polysemantic approach; analysis of the modals makes clear that both categorical and non-categorical approaches are relevant and therefore an adequate description of the meanings of modals must achieve a synthesis of these two approaches.” (ibid:10)
21
can be explained as derivatives of non-epistemic uses8: while non-epistemic uses
relate directly to their according action processes, epistemic uses relate to the mental
processes that are the precursor of these action processes. Brinkmann (1971) also
proposes a classification of modal verbs based on the notion of actions (be they
physical or mental). His attempt to describe modal verbs in terms of general
communicative functions is fruitful because it arrives at more abstract and systematic
categorisations. Brinkmann‟s framework will be discussed in more detail in section
2.2.2 as the basis for the current investigation.
Brünner and Redder‟s (1983) approach is also compatible with another framework
that proposes one basic meaning for each modal. This is Kratzer's (e.g. 1978, 1991,
forthcoming) account of modality within a Mögliche-Welt-Semantik (possible-world
semantics) framework. Here, modal verbs have a constant, basic meaning and the
different interpretations of the modal verbs are determined by varying
Redehintergründe (discourse backgrounds) – i.e. information supplied through the
utterance situation – which propose a deontic, epistemic etc. reading on the
background of a „possible world‟, in which the suggested proposition is true. This
framework is also adopted for the account of modality in the "Grammatik der
deutschen Sprache" (Zifonun et al. 1997) produced by the Institut für deutsche
Sprache (IDS), which constitutes a modern and one of the most comprehensive
descriptive grammars of the German language: “For each modal verb we therefore
assume one meaning and one syntactic category. This one meaning usually entails
several types of informational backgrounds within one modal relation” (emphasis in
original)9.
All of these approaches advocate the view that the meaning of modal verbs is
entirely dependent on context. This context can be provided by the immediate
surrounding discourse, but Nehls (1986:5) demonstrates that even very broad
contexts such as the speaker‟s or hearer‟s „experience of life‟ can influence how a
modal verb is interpreted: Whereas MUST in the sentence
(3) You must be drunk.
would, in most circumstances, be interpreted epistemically as a confident conclusion
of the stated proposition, MUST in the sentence
8 According to some linguists, rather than through metaphorical processes this derivation is achieved through the conventionalisation of “context-induced reinterpretation” (Heine, Claudi and Hünnemeyer 1991) where specific contexts or circumstances invite new inferences or converstational implicatures and ultimately lead to the emergence of new grammatical meanings. (See also e.g. Traugott and König 1991.)
9 "Wir gehen somit für jedes Modalverb von einer Bedeutung und einer syntaktischen Kategorie aus. Diese jeweils eine Bedeutung umfaßt in der Regel mehrere Arten von Redehintergründen unter einer modalen Relation."
22
(4) You must be sober.
would probably rather be interpreted deontically in the sense of a strong obligation.
2.1.3 Other types of modality and other means of expressing modality
It has to be noted at this point that the majority of the earlier theoretical approaches to
modality concentrated almost exclusively on modal verbs. Öhlschläger (1984:230)
lists several reasons for this preoccupation (and while he only discusses work on
German modality, the same reasons hold true for English). Amongst these are their
unique morphological status (see 2.3.1.1) and etymology, their semantic ambiguity,
but above all the fact that despite their vast variety of meanings they seem to form
some kind of closed semantic system. It is not surprising then that the most influential
work on German modality started out with models of the semantic and paradigmatic
relationships between modal verbs (e.g. Bech 1949) and descriptions of their
variations in function and meaning (e.g. Welke 1965). Even later models of modality
that are based on definitions of modality explicitly not restricted to modal verbs, treat
them as the core repertoire of modality and base their models on them. For example,
Calbert (1975:1) sets out to develop a logico-semantic system of modality as a model
for all utterances with expressions that convey “various degrees of „possibility‟ and
„necessity‟”, but then is again mainly concerned with modal verbs. This focus on
modal verbs, however, brings with it two fundamental issues.
First of all, when investigating modal verbs one cannot fail to realise that they can
be used to express more than epistemic and deontic meanings in the sense
discussed so far. In German, modal verbs are not only used to express speaker
commitment in terms of possibility and necessity (the logical definition of epistemic),
but also in terms of what kind of evidence there is for a proposition („quotative‟). The
modal verbs SOLLEN and WOLLEN can be used for so called „evidentials‟ to express
either what someone other than the speaker or the subject says:
(5) Er soll sich eine Luxusvilla in Monaco gekauft haben. „He is said to have bought a luxury villa in Monaco.‟10
or to express what the grammatical subject of the sentence claims:
(6) Er will auf der Hochzeit der Königin gewesen sein. „He claims to have been to the Queen‟s wedding.‟
Another kind of modality involves the modal verb KÖNNEN. In the sentence
10
In English, evidentials are not grammaticalised in modal verbs.
23
(7) Er kann sehr gut Klavier spielen. „He can play the piano very well.‟
the modal verb expresses a meaning of „having the ability or the mental/physical etc.
disposition to...‟, which is neither epistemic nor deontic, but what has been termed
„dynamic‟ modality (cf. von Wright 1951:28, Palmer 1979:7). There is therefore again
considerable discussion whether everything that can be expressed by modal verbs,
including evidential and dynamic modality, should be subsumed under the term
modality (a view that e.g. Welke (1965), Brinkmann (1971), Zifonun et al. (1997)
embrace, but Öhlschläger (1984) opposes).
Some linguists, however, identify even more types of modality. These other types
usually arise when modality is defined very broadly, as e.g. by Rescher (1968:24-6):
“A proposition is presented by a complete, self-contained statement which, taken as a
whole, will be true or false […]. When such a proposition is itself made subject to
some further qualification of such kind that the entire resulting complex is itself once
again a proposition, then this qualification is said to represent a modality to which the
original proposition is subjected.” This (too) broad definition (cf. Palmer 1986:12)
leads Rescher to identify not only alethic, epistemic and deontic modality, but
additionally also
„temporal‟ (It is sometimes/often/always the case...)
„boulomaic‟11 (It is hoped/feared/regretted/desired that...)
„evaluative‟ (It is a good/ wonderful/bad/horrible thing that...)
„Causal‟ (The state of affairs will bring it about/prevent (impede) its coming about
that...)
and three types of „conditional‟ modalities.
The problem with this definition and is that one of Rescher‟s types of modality can be
applied to virtually every complex sentence, which neither helps to clarify the nature
of modality nor to identify a set of linguistic expressions of modality.
On the other hand, Rescher‟s account at least considers other lexical expressions
that can convey modal meanings and de-focuses modal verbs. And this is the second
of the two issues arising from concentrating on modal verbs: the negligence of other
means to express modality. In empirical research, the restriction to modal verbs is
sometimes justified by the purpose of the study. Coates (1983), for example,
specifically sets out to use corpus analysis in order to develop a model of “The
11
This is usually referred to as „volitives‟.
24
Semantics of the Modal Auxiliaries” in English and thus seeks to contribute to a
clearer picture of the meaning relationships between, specifically, modal verbs.
(Although she recognises paraphrases of modal verbs, such as necessary for,
advisable for, perhaps, possible that etc., they are not included in her analysis.)
Palmer (1979:1) on the other hand, restricts his account of “Modality and the English
Modals” to modal verbs simply by stating that “there is a grammatical, or semantic-
grammatical, category called modality and [...] it is expressed in English by the modal
verbs”, which results in a rather unjustified exclusion of any other means of modal
expression.
More recent work on modality, and especially corpus studies, do, however,
acknowledge that modal meanings are quite commonly expressed by a variety of
modal devices besides modal verbs and incorporate those in their investigations (for
English cf. for example Perkins 1983, Holmes 1988, Hoye 1997, Biber et al. 1999,
Aijmer 2002; for German cf. for example Wichmann and Nielsen 2000).
For German, Öhlschläger (1984:231) specifically identifies it as a "Vorteil"
(advantage), that e.g. Calbert's system of modality is not restricted to modal verbs but
includes other modal devices, for instance lexical verbs such as verlangen,
wünschen, erlauben behaupten, annehmen etc. (demand, wish, allow, claim, assume
etc.). From a contrastive point of view, Nehls (1986:176ff) is able to empirically
underpin Edmondson et al.'s (1977:256) remark that, generally speaking, German
epistemic modality is more often than in English expressed through modal adverbs
such as sicherlich, wahrscheinlich, vielleicht (surely, probably, perhaps) or other
verbal, adjectival or nominal constructions rather than through modal verbs. For
English, the opposite is true as the modal verb occupies a much more central
position, especially in epistemic modality (cf. Leech and Svartvik 1975:292). Perkins
(1983:104) argues that this is the case because the modal verbs are more integrated
into the grammatical and syntactic system of the English sentence and are less
specialised in their meaning than other modal expressions.
2.1.4 Terminological differences
The debate over how to define modality and what to include as modal expressions is
far from resolved. However, there seems to be a general consensus to keep a
dichotomous distinction between epistemic and non-epistemic modality. In this
25
dichotomy, evidentials are usually subsumed under epistemic modality12, whereas
other types of modality such as deontic, dynamic or existential are grouped together
as non-epistemic modality. This is also sometimes referred to as root modality.
Coates makes the point that trying to separate different non-epistemic meanings into
discrete categories (like deontic, dynamic etc.) leads to “arbitrary cut-off points” that
“obscure the essential unity of the Root modals” (1983:21).
In general, the choice of categories and most of all terminology is strongly
dependent on the theoretical framework that the respective researchers adopt.
Realising that – unlike modality in the logical sense – “the overall picture of the
modals is extremely messy and untidy” (Palmer 1979:40), many linguists have turned
to other approaches and frameworks to address the concept of modality. This leads
to a variety of promising approaches to modality but also to a confusing wealth of
categorisations and terminology (for a recent overview of theories and terminology
see also Droessinger 2004). The following (far from exhaustive) list should illustrate
the diversity of terms that have been used just for these two basic categories of
epistemic and non-epistemic modality. The differences in terminology are simply
grounded in the fact that modality is explored within different theoretical frameworks:
12
Cf. however Fagan (2001) for an argument that the evidential functions of the German modal verbs WOLLEN and SOLLEN cannot be included under epistemic modality.
26
Linguists advocating this terminology
epistemic modality non-epistemic modality
Coates (1983), Sweetser (1982)
epistemic root
Calbert and Vater (1975), Brünner and Redder (1983)
inferential
non-inferential
Helbig and Buscha (2001), Palmer (1986)13
subjective objective
Palmer (1990) epistemic discourse-oriented (subject-oriented for dynamic modality)
Quirk et al. (1985) extrinsic intrinsic
Bybee and Fleischmann (1995), Heine (1995)
epistemic agent-oriented
Nehls (1986) wahrscheinlichkeitsorientiert
(probability oriented)
handlungsorientiert (action oriented)
Brinkmann (1971) Bedingungen für die Geltung einer Information (conditions for the validity of information)
Voraussetzungen für eine Realisierung (conditions for an implementation)
Diewald (1999) deiktisch (deictic)
nicht-deiktisch (non-deictic)
TABLE 2: OVERVIEW OF TERMINOLOGY FOR EPISTEMIC AND NON-EPISTEMIC MODALITY
In summary, we can determine that there is a common core in the linguistic
definitions of modality – modality characterises the relationship between speaker and
predication and between predication and reality (Augustin 2006:13). Despite this,
however, it is still widely disputed in the vast amount of thorough explorations of
modality within various theoretical frameworks how modality should be categorised
and which linguistic means count as modal expressions. This wide range of different
theoretical positions, schools of thought and sometimes language-specific linguistic
traditions is reflected in a whole jungle of terminology that has become almost
impossible to trace. The discussion is ongoing and far from resolved, so before an
investigation into modality in learner language can begin, two fundamental points
need to be determined:
13
N.B. These terms are particularly confusing as in some accounts the subjective/objective distinction is a simple replacement of the terms epistemic/non-epistemic (Helbig and Buscha 2001, Dreyer and Schmitt 2000). In other models, however, subjectivity/objectivity is a subdivision of epistemic/non-epistemic modality, so there can be instances of subjective non-epistemic and objective epistemic modality (e.g. Coates 1983, Palmer 1986).
27
1. Which theoretical approach to modality provides a sound basis for the investigation
of learner modality and which types of modality should be included?
2. Which types of modality does the selected approach entail and which linguistic
expressions of these types are to be included in the investigation?
The following sections are dedicated to answering these questions and presenting
the model of modality that has been chosen as the theoretical basis for the present
study and thus to developing a detailed picture of the object of investigation.
2.2 MODALITY IN THE GERMAN LANGUAGE
From the previous section it has become clear that the concept of modality is rather
elusive and 'unwilling' to lend itself to any clear-cut definitions or categorisations. A
wide variety of approaches with different foci have been pointed out and all seem to
have some stronger as well as weaker points with regards to their suitability for the
present investigation, be it because of underlying conceptual or practical reasons.
Palmer (1986:6) stresses that it is “often very difficult to decide what to include and
what to exclude from a grammatical study of modality” as “both semantic
considerations and judgements about grammaticality have to be made
simultaneously”. As this investigation is less concerned with the grammatical study of
modality but rather with the interlanguage modal system of learners, the deciding
factors for an appropriate theoretical framework and definition of modality must be
determined by the purpose of the study.
2.2.1 Requirements of a theoretical approach for the investigation into learner modality
The present study investigates usage patterns of modality features by second
language learners. This means that the account of modality that is chosen for this
investigation has to define modality and modal expressions in an integrative way that
provides a sound theoretical basis for the production of meaningful information about
how second language learners use modal expressions in comparison to native
speakers and how these patterns change with advancing proficiency. The term „use‟
here includes e.g. counts and comparisons of frequencies of certain grammatical and
lexical items, but it goes beyond that. The nature of the corpus data is such that it
allows for more detailed analysis of differences and distributions with regard to
various semantic levels as well as different contextual factors. An appropriate model
28
of modality must therefore provide systematic categorisations of the semantic
functions of modal expressions. As this analysis aims to provide information that
should, eventually, benefit the learners in their acquisition of a native speaker-like
system of modality, the approach to modality in learner language has to start out with
the learner‟s needs in this particular area. On the one hand, language learners have
to come to terms with the morphological and grammatical features of modal
expressions, but beyond that they also have to master the semantic and pragmatic
system of modality and with it its various lexical manifestations in the target language
in order to adequately express modal meanings. And while the morphological and
grammatical rules for expressing various kinds of modality are already relatively
complicated, the actual use of these expressions in terms of their semantic and
pragmatic meaning, but also in terms of their relative frequencies and contextual
restraints, is even more intricate and complex.
At the same time, expressions of modality are high-frequency phenomena in
German, so learners are confronted with them early on in the learning process and
will develop an extensive and complex interlanguage system for modality over a long
period of time. For these reasons, it is proposed for the present study to adopt a
relatively broad research scope that allows taking into account the full breadth of
modal means that learners are presented with and have to deal with in the course of
their foreign language acquisition14. This means that, on the one hand, an account of
modality is chosen that is form-based in a way that it includes all different meanings
of modal verbs, even those that some researchers exclude as not modal in a stricter
sense (e.g. dynamic use of KÖNNEN, Öhlschläger 1984:242). On the other hand, an
account of modality is needed that is content-based in a way that it includes all the
different linguistic means to express modality (that is to say, not only modal verbs, but
also modal adverbials and other lexical expressions) and categorises them according
to abstract and systematic semantic criteria.
This is particularly important as differences in semantic categories across
languages can lead to confusions and difficulties with certain modal items. As Nehls
(1986:5), in his comprehensive contrastive analysis of English and German modal
verbs, states: “The author knows from his own experience, teaching at university
level, how persistent errors involving modal verbs are, from first year right through to
final exams. These errors usually stem from two sources: L1 German interferences
14
It has to be noted, however, that I follow the position of Kątny (1993:42) here, who views modality as "an optional category of the sentence" where the indicative is "unmarked in terms of modality". This study will therefore concentrate on utterances that display some explicit form of lexical or grammatical marker of modality.
29
and confusions within the English modal verb system, which can be traced back to
the inability to semantically discriminate between different modal meanings on the
part of the learners.”15 And there is no reason why the same should not be true in the
opposite direction, i.e. for L1 English speakers learning German. In addition to
employing an inclusive definition of modality, contrastive information on the German
and English modal system will therefore be drawn upon in order to cast as wide a net
as possible to uncover patterns and potential problems in the learners‟ interlanguage
systems of modality. The following sections will explain concretely which items are
included in the analysis on the basis of the theoretical framework underlying this
study. However, as Palmer (1986:6) remarks concisely, “some degree of arbitrariness
in the choice of items is inevitable.”
2.2.2 Brinkmann‟s account of the modal system
An account of modality that offers the kind of abstract semantic categorisation at the
same time as comprehensiveness in terms of linguistic means which have been
defined as the two key components for a suitable theoretical basis for the present
study is provided by Brinkmann (1971).
Brinkmann's account is born out of the Inhaltsbezogene Grammatik16, a
specifically German development of structuralism theories based on the linguistic
philosophy of Wilhelm v. Humboldt17. In this theory, language is taken as a formative
activity of the human mind that constitutes a Zwischenwelt (in-between world)
between reality and thinking. The approach to grammar is holistic insofar as it is
concerned with the exploration of these linguistic Zwischenwelten as mediating
instances between human beings and reality, which is captured and structured by
language.
Unlike other accounts of modality in grammatical and lexicological volumes that try
to list all the different meaning variants of modal expressions (cf. for example Schulz
and Griesbach 1955, Helbig and Buscha 2001), Brinkmann centres his approach to
modality around the abstract relations that modal expressions establish. Modality in
15
"Der Verfasser weiß aus seiner eigenen Erfahrung im Universitätsunterricht, wie hartnäckig sich die Fehler im Bereich der Modalverben vom ersten Semester bis hin zum Abschlußexamen erweisen. Dabei liegen die Fehlerquellen einerseits in der Interferenz der deutschen Muttersprache, andererseits aber auch in Verwechslungen innerhalb des englischen Modalverbsystems, die auf ein zu geringes semantisches Unterscheidungsvermögen in diesem Bereich zurückzuführen sind."
16 Content-based grammar, also called Sprachinhaltsforschung (language content research), developed in the 1950s by Leo Weisgerber.
17 Note, however, that influences of these theories can also be traced in anthropological metalinguistic theories outside Europe, especially in the work of Benjamin Whorf (1956) and Edward Sapir (1958) (“principle of linguistic relativity“) and George Lakoff (1987).
30
this theory is defined from a communicative functional point of view, which on the one
hand leads to more systematic meaning categories, and on the other hand allows for
the inclusion of a variety of linguistic expressions. He begins his account of the modal
system with the following definition:
“Modality is defined as the validity that is linguistically assigned to an utterance.”18
(Brinkmann 1971:357).
He goes on to ascertain that the interpretation of this validity depends on the
communicative aim:
“How the validity is to be understood is expressed through different processes,
which operate on two different levels – two levels of modality. An utterance is
either designed to change reality or to grasp reality. The former is concerned
with reality, or more precisely: an implementation, the latter with truth. The finite
verb formulates either conditions for the implementation or conditions for the
validity of information.”19 [my emphasis] (Brinkmann 1971:359)
These two levels correspond loosely to the non-epistemic/epistemic distinction, but
the crucial point here is that modal expressions are not described in terms of „basic
meanings‟ but in terms of abstract relations (cf. Brünner and Redder 1983:36). To
illustrate this point, Brinkmann (1971:359f) gives the following example:
(8) Er soll Urlaub machen.
He determines that the fundamental function of the modal verb SOLLEN is to signal
the involvement of an external authority or intermediary agent other than the subject
or the speaker. How the role of this external authority functions within a concrete
utterance, however, is not dependent on the modal verb SOLLEN, but on the type of
communication. On the first level of modality, the speaker (e.g. the boss of a
company) addresses the intermediary agent in order to achieve the desired
implementation („Tell him to go on holiday‟). On the second level of modality, the
speaker passes on a piece of information that he has received from the (in the case
of SOLLEN unspecified) external agent („He is said to be on holiday‟). The modal
verb itself only denotes a general aspect of meaning, which is only concretised
18
"Unter Modalität wird die Geltung verstanden, die einer Äußerung sprachlich zuerkannt wird." 19
“Wie die Geltung gemeint ist, bringen andere Verfahrensweisen zum Ausdruck, die auf zwei verschiedenen Ebenen operieren, zwei Ebenen der Modalität. [...] Entweder ist eine sprachliche Äußerung darauf angelegt, die Wirklichkeit zu ändern [...] oder die Wirklichkeit zu erfassen [...]. Im ersten Fall kommt es auf die Wirklichkeit, genauer: eine Realisierung an, im zweiten Fall auf die Wahrheit. Die Personalform des Verbums formuliert entweder Voraussetzungen für die Realisierung oder Bedingungen für die Geltung einer Information."
31
through the connection with a specific infinite verb and through its employment in a
specific situation – they are entirely context dependent.
Brinkmann (1971:361) shows that apart from modal verbs there are three other
types of linguistic means that can fulfil the function of assigning an assessment of the
validity to an utterance, so that he ends up with a list of four linguistic options to
express modality:
1. Modal verbs (KÖNNEN, MÜSSEN, WOLLEN, SOLLEN, MÖGEN, DÜRFEN,
WERDEN)
2. Modal adverbs (e.g. möglicherweise, vielleicht, sicherlich)
3. Modal infinitives (ist … zu, hat … zu)
4. Mood (subjunctive)
These four options will be described in detail in the following section. First,
however, some further general remarks have to be added. Brinkman (1971:362)
picks out modal verbs and subjunctive mood as the most important means for
shaping the modal system, probably because they are the two most flexible and
unspecified means for expressing modality20. In his further explorations of the modal
system, he therefore concentrates on these two means and only makes brief
comments about modal adverbs and modal infinitives. In order to obtain a more
comprehensive picture of all of the modal means, I will draw upon Zifonun et al.
(1997) for additional information. Their “Grammatik der Deutschen Sprache” is an
attempt at an academic German grammar that describes the current state of the art in
German grammar without actually adopting just one theoretical perspective. In the
different parts of the grammar they attempt to develop the complementary effects of
functional and form-based theories. Although they base their model of modality on
different linguistic theories (for the semantic analysis mainly Kratzer‟s (1991)
Mögliche-Welt-Semantik) and use different terminology, their approach displays
some fundamental similarities to Brinkmann‟s. The most prominent one is that
modality is viewed as a concept that establishes through discourse how the validity
(Geltung) of a proposition is to be understood. By introducing a modal element into
an utterance,
“the speaker does not directly claim the truth of a proposition within his/her
understanding of the world, but he/she relates this proposition to a specific
discourse background. These discourse backgrounds are part of the text- or
discourse knowledge: They are introduced in texts or discourses by the 20
“Am wichtigsten für die Gestaltung der Modalität sind von diesen vier Möglichkeiten der Modus des Verbums und das Modalverb.“
32
author or speaker, are perceived by the addressees or participants in an
interaction and are dynamically altered in the course of the interaction.”21
(Zifonun et al. 1997:1882) [emphasis in original]
A second similarity between Brinkmann and Zifonun et al. is that both accounts do
not pertain solely to modal verbs, but what Brinkmann calls the whole „modal system‟.
While Brinkmann, however, focuses on modal verbs and mood, Zifonun et al.
(1997:1884) stress that "the concept of the discourse background is not only to be
applied to the description of modal verbs, but also to other modal expressions,
especially the modal adverbials.”22. This is an important point in the pursuit of a
comprehensive study of modality in learner language. At the same time, they point
out the semantic similarity (N.B. not synonymy) of modal adverbials and modal verbs
(Zifonun et al. 1997:1260), which again underlines the goal of comprehensiveness in
their approach:
(9) Er hat vielleicht/sicher einen guten Grund gehabt, nicht zu kommen. He has maybe/surely a good reason had not to come „Maybe/Surely he had a good reason not to come.‟
(10) Er kann/muss einen guten Grund gehabt haben, nicht zu kommen. He can/must a good reason have had not to come „He may/must have had a good reason not to come.‟
So far, we could establish that Brinkmann (1971) – supplemented with additional
information from Zifonun et al. (1997) – offers an account of the German modal
system that can serve as a fruitful basis for the investigation of learner modality as it
provides both an abstract and systematic semantic description of modal meanings
and inclusivity in terms of linguistic expressions. In the following section, the formal
and semantic properties of the four types of modal means that have been introduced
in this section will be discussed in more detail.
21
“…ein Sprecher nicht direkt einen Wahrheitsanspruch für den thematisierten Sachverhaltsentwurf gegenüber der Welt, wie er sie sieht, anmeldet, sondern daß er diesen Sachverhaltsentwurf auf einen speziellen Redehintergrund bezieht. Einzelne Redehintergründe sind Teile des Text- oder Diksurswissens: Sie werden in Texten oder Diskursen durch den Autro bzw. Sprecher eingebracht, durch die Adressaten oder Teilnehmer einer Interaktion wahrgenommen und im weiteren Verlauf dynamisch verändert.“
22 “[d]as Konzept des Redehintergrundes ist im übrigen nicht nur zur Beschreibung der Modalverben, sondern auch anderer Modalwörter, vor allem der modalen Adverbialia vorzusehen.“
33
2.3 LINGUISTIC MEANS FOR EXPRESSING MODALITY
In this section, the four linguistic means of expressing modality in German will be
presented in more detail. The focus of this description will be a summarising
categorisation of their semantic properties, but formal aspects will also be included
where necessary and appropriate. Unless otherwise indicated, the following details
are based on Brinkmann (1971) and Zifonun et al. (1997).
2.3.1 Modal verbs
The German modal verbs are MÜSSEN, KÖNNEN, WOLLEN, SOLLEN, DÜRFEN,
MÖGEN. All of these verbs are available for both levels of modality, conditions for
implementation (non-epistemic modality) and validity of information (epistemic
modality), but they display different formal and grammatical restrictions within these
two types. It is therefore necessary to introduce some formal characteristics of modal
verbs first, before the semantic categorisation according to Brinkmann (1971) is
presented. A case will be made later on for the inclusion of WERDEN, which is only
available for epistemic modality.
2.3.1.1 Formal characteristics
Morphologically and syntactically modal verbs exhibit the following characteristics23
(examples from Zifonun et al. (1997:1255ff)):
1. They govern the bare infinitive (without zu).
(11) Ich muss gehen. „I must go.‟
2. If they are not used with an infinite verb they can govern a directional complement:
(12) Ich will nach Hause. I want to home „I want to go home.‟
3. The modal verbs WOLLEN (want), MÖGEN (like) and KÖNNEN (can) can also act
as main verbs and take an accusative complement instead of the infinite verb24:
23
N.B. this list is not entirely exhaustive (for a comprehensive illustration see Zifonun et al. 1997:1255ff) but represents a summary of the most prominent features of modal verbs also with a view to differences to the English modal verb system (for the morphological and syntactic differences between German and English modal verbs see Nehls 1986:17ff); each individual characteristic does not distinguish modal verbs from all other verbs and some of the characteristics are typical for but not exclusively inherent to modal verbs. Taken together, however, and in light of the unique semantic characteristics of modal verbs, the list excludes comprehensively other main and auxiliary verbs.
34
(13) Ich will ein Eis. „I want an ice-cream.‟
(14) Ich mag ihn. „I like him.‟
(15) Ich kann Englisch I can English „I can speak English.‟
4. They have no imperative form.
5. They do not form a passive, although they can govern a passivised infinitive.
(16) Die Rechnung muss heute bezahlt werden. „The invoice must be paid today.‟
6. If the modal verb governs an infinitive, the past participle of the modal verb is
replaced by the infinitive:
(17) Er hat das Gedicht nicht aufsagen können. He had the poem not recite.INF can.INF „He couldn‟t recite the poem.‟
The past participle is only used if the modal verb is used as a main verb:
(18) Er hat das Gedicht nicht gekonnt. He had the poem not can.PP „He couldn‟t recite the poem.‟
7. More than one modal verb can occur within the same clause:
(19) Sie muss kommen können. She must come can.INF „She must be able to come.‟
8. Conjugational particularities:
In the present tense, 1st and 3rd person singular have Ø-ending:
ich/er muss – kann – mag – darf – will – soll.
KÖNNEN, DÜRFEN, MÖGEN, WOLLEN change the stem vowel between singular
and plural forms: ich kann – wir können; ich darf – wir dürfen; ich mag – wir mögen;
ich will – wir wollen;
The preterite is formed from the plural stem without umlauts in weak conjugation:
ich konnte – ich durfte – ich mochte – ich wollte – ich musste – ich sollte
The past subjunctive (Konjunktiv II25) forms of DÜRFEN, KÖNNEN and MÜSSEN
keep the umlaut (ich dürfte/könnte/müsste), SOLLEN and WOLLEN have no umlaut
(ich sollte/wollte) and are therefore identical with the preterite indicative forms.
24
Brinkmann (1971:382) explains this conclusively with the fact that WOLLEN, MÖGEN and KÖNNEN imply that the grammatical subject is the agent that can influence its environment. SOLLEN, MÜSSEN and DÜRFEN, on the other hand, imply that the grammatical subject is under the influence of an external agent. The subject is therefore not able to act as agent.
25 The terms Konjunktiv I and Konjunktiv II will be used throughout this study to refer to the two types of German subjunctive mood, which are also often called Konjunktiv Präsens (present subjunctive) and
35
Zifonun et al. (1997:1884) emphasise that there is a clear 'caesura' between
epistemic and all the other (non-epistemic) discourse backgrounds. Apart from the
obvious semantic differences, this is manifest in the fact that the modal verbs exhibit
different grammatical constraints and circumstances depending on which type of
modality they express (Zifonun et al 1997:1268ff):
The most prominent of these is that perfect and future tense paradigms are
restricted for modal verbs in epistemic use. The following sentences can only have
non-epistemic reading:
(20) Sie hat ihre Schulden zahlen müssen. She has.PAST her debts pay must.INF 'She had to pay her debts.'
For an epistemic reading, the perfect tense form of the modal verb is not available,
i.e. to achieve a perfective reading for an epistemic utterance, the infinitive has to
carry the perfect tense marking:
(21) Sie muss ihre Schulden bezahlt haben. She must her debts pay.PP have 'She must have paid her debts.'
Other restrictions include e.g. that DÜRFEN is only available in epistemic sense in
the Konjunktiv II, indicative forms (present tense or preterite) exclude an epistemic
reading:
(22) a) Das Paket dürfte inzwischen in Sydney angekommen sein. The parcel DÜRFEN-KONJII by now in Sydney arrived be 'The parcel should have arrived in Sydney by now.' b) *Das Paket darf inzwischen in Syndey angekommen sein. The parcel DÜRFEN-PRES by now in Sydney arrived be‟ *Das Paket durfte inzwischen in Syndey angekommen sein. The parcel DÜRFEN-PRET by now in Sydney arrived be‟
Furthermore, there is a syntactic restriction on epistemically used modal verbs in
that in combinations with other modal verbs they can only occur as the finite part of
the verbal complex. MUSS in example (19) Sie muss kommen können. above carries
epistemic meaning, KÖNNEN in this example cannot be interpreted epistemically.
Other uses, e.g. with an accusative complement (23), directional complement (24)
or subordinate clause (25) are not possible in epistemic senses, neither are
nominalisations (26), so all these examples have to be interpreted non-epistemically:
Konjunktiv Präteritum or Imperfekt (past subjunctive). For a more detailed account of the subjunctive mood in German see 2.3.5.
36
(23) Das kleine Mädchen will ein Eis. The little girls wants an ice 'The little girl wants an ice-cream.'
(24) Ich muss weg. I must away 'I have to go.'
(25) Ich möchte, dass du ihm hilfst. I would like that you him help 'I would like you to help him.'
(26) Das Wollen allein nützt nichts, Können ist gefragt. The want-NOM alone uses nothing, can-NOM is asked 'Wanting alone is useless, ability is called for.'
On top of these grammatical/syntactical conditions, which can prohibit certain
interpretations, there are other factors that can affect how a modal verb is interpreted,
which leads to different distributional patterns for the two types of modality. Research
into this area benefits immensely from corpus linguistic methods, which enable the
researcher to establish „typical‟ usage patterns through detailed frequency analyses.
Heine (1995) for example examines, by way of a corpus investigation, factors that
contribute to an utterance being associated with one or the other kind of modality. He
identifies the following three groups of factors (Heine 1995:25ff):
The first one pertains to the linguistic context: Non-epistemic readings are very
likely if the main verb is an action or terminative verb, if the modal verb is in the
perfect or preterite tense, if it occurs in interrogative or negative utterances and if
the subject is first or second (rather than third) person. Conversely, epistemic
senses are more likely if the main verb is in the perfect tense or is a stative verb.
Secondly, however, linguistic categorisations alone can often not distinguish
epistemic from non-epistemic meanings. “What has to be taken into consideration
in addition are the contextual frames with which particular uses of a modal are
associated”(Heine 1995:47). Contextual frames, also called „inferential schemata‟
or „frames‟ are defined as “a body of knowledge evoked by the language user in
order to provide an inferential basis for the understanding of an utterance” (Heine
1995:27f). This knowledge can be based on contextual clues, knowledge of the
world, social norms, but possibly also involve factors like frequency of experience,
perceptual salience and stereotypes (and may be roughly equated to Kratzer‟s
(1991) discourse backgrounds). These contextual frames can ultimately override
the expectations evoked by the linguistic context and, for example, discourage a
37
non-epistemic reading even in an utterance with a human agent as subject and a
dynamic main verb.
Thirdly, the most important factor in determining which variety of modality a given
modal utterance receives is, according to Heine, the “presence vs. absence of the
modal force”. This force is characterised by an element of will, i.e. that there is an
interest in an event either occurring or not occurring. Instances of epistemic
modality typically lack exactly this conceptual property.
Although in most instances these factors will produce one or the other reading as
favoured, there are cases that are indeterminate between epistemic and non-
epistemic senses. Coates (1984:15ff) investigates this indeterminacy in detail for the
English modal auxiliaries and establishes three different types: gradience, ambiguity
and merger. Gradience concerns the nature of the continuum of meaning within a
given modal category (epistemic or non-epistemic), for example from the core of
ABILITY to the periphery of POSSIBLITY of English can. Ambiguity concerns
expressions that have two different senses in discrete modal categories and it is not
possible to decide which of these is intended (either/or relationship). Merger also
refers to expressions that have two senses in discrete categories, but they are in a
both/and relationship, i.e. they are mutually compatible. It is not necessary to decide
which of the two senses is intended in order to understand the utterance.
Zifonun et al. (1997), Heine (1995) and Diewald (1999) all agree that the
explanation for the grammatically, syntactically and functionally divergent behaviour
of modal verbs in different meanings is to be found in the course of their
grammaticalisation or, in Diewald's words (1999:54), in the fact that “the structural
options for modal verbs correlate strongly with their different usages, which display
varying degrees of grammaticalisation”26. Diewald (1999:27ff) gives a detailed
account of the proof of the proposed course of grammaticalisation which comes to
the conclusion that what she calls „deictic‟ (epistemic) uses of modal verbs are more
grammaticalised than „non-deictic‟ (non-epistemic) uses. It will be shown in the
discussion of results later on (see 6.2.4) that the varying degree of
grammaticalisation of the different modal verb uses might contribute to explaining
why certain uses seem more difficult for learners to acquire than others.
26
“die strukturellen Möglichkeiten der Modalverben direkt mit ihren unterschiedlich stark grammatikalisierten Gebrauchsweisen korrelieren”
38
2.3.1.2 Semantic characteristics
As discussed in the previous chapter, the semantic characteristics of modal verbs will
be explained in terms of a network of abstract relations according to Brinkmann
(1971). He distinguishes two fundamental types of relations that are relevant for
describing modality. These two types define two groups of modal verbs (for detailed
and exemplified descriptions see individual modal verbs below).
In the first group, we encounter modal verbs “that give the subject an explicit
direction for the execution of the process conveyed by the infinite verb”27 (Brinkmann
1971:386). The modal verbs in this group are WOLLEN, SOLLEN and DÜRFEN. The
difference between them is the source of the authority that gives the subject the
direction for execution. In the case of WOLLEN, the authority lies with the
grammatical subject; SOLLEN, as exemplified above, implies an external authority
that gives the direction for execution to the subject; WOLLEN and SOLLEN can
therefore depict converse perspectives of the same proposition:
(27) Ich soll einkaufen gehen = Jemand will, dass ich einkaufen gehe. „I am supposed to go shopping.‟ = „Somebody wants me to go shopping.‟
With DÜRFEN, the subject itself gives the direction for execution (as with WOLLEN),
but an external authority is required that legitimises this direction.
In the second group, the modal verbs do not determine the direction for an
execution, but rather the nature of the circumstances pertaining to the process
conveyed by the main verb. The modal verbs in this group are KÖNNEN and
MÜSSEN. KÖNNEN implies that the circumstances are viewed as such that an
implementation is possible or conceivable, whereas for MÜSSEN the circumstances
are viewed as making an implementation compulsory.
MÖGEN occupies a special position in this classification as it has parts in both
groups. While the indicative form belongs to the latter group in that it acknowledges
that the circumstances make an implementation possible (like KÖNNEN), the
Konjunktiv II form MÖCHTE has almost become a modal verb in its own right that
belongs to the former group as it expresses the same kind of direction for an
execution originating from the subject as WOLLEN28.
These two fundamental relational types hold true on both levels of modality
identified in the previous section. Each modal verb, therefore, conveys a unique
27
“…, die dem Subjekt eine ausdrückliche Richtung auf den Vollzug des im Verbum (durch Infinit I) genannten Prozesses zuschreiben;”
28 For the analysis of MÖCHTE as independent verb see also Ehlich and Rehbein (1972:318).
39
fundamental relation that defines either the characteristics of the conditions for an
implementation or the characteristics of the conditions for the validity of information. It
is these relational types that give the different modal verbs their different meanings;
what other accounts list as different usages of the same modal verb can in
Brinkmann‟s modal system be explained from just this one basic relation.
2.3.1.2.1 MÜSSEN
MÜSSEN implies that the speaker acknowledges that a necessity exists for the
proposition conveyed in the utterance. Applied to the two levels of modality this
means that this necessity either pertains to an implementation, i.e. the speaker
acknowledges that the conditions are such that there is a necessity to do something,
as in
(28) Die Rechnung muss heute bezahlt werden. „The invoice has to be paid today.‟
The necessity can also pertain to the validity of information, i.e. the speaker asserts
that the conditions are such that there is a necessity to assume the given proposition
as true, as in
(29) Der Brief muss gestern abgeschickt worden sein. „The letter must have been posted yesterday.‟
We are therefore either dealing with some form of duty or compulsion or a compelling
conclusion that excludes other possibilities. The reasons why the given conditions are
viewed as compulsory are manifold. When dealing with implementations, they can lie
in human nature or social/economical/legal etc. rules and norms or the specific
circumstances of a particular situation or they can be necessary prerequisites in order
to achieve a certain goal. When dealing with information they lie in the speaker‟s
knowledge of the world or their knowledge of how certain processes typically take
place.
The Konjunktiv II form MÜSSTE does not imply a different value from MÜSSEN
(like e.g. DÜRFTE, see below). It merely modifies the indicative in the way that is
characteristic for Konjunktiv II in general, namely that it shifts the proposition
conveyed in the utterance beyond the given horizon29, which makes the assumption
or duty more tentative or attenuated (see also Diewald 1993):
29
Brinkmann introduces the term Horizont (1962:105) first in works on the relation between dialect and standard language. Later, he transfers this idea to communication in general where conversational partners meet under the influence of their respective „horizons‟: “This includes their relationship to each other, their social roles, their experiences, memories and expectations, of course also all their previous knowledge, everything they think about.” (“Es schließt ihr Verhältnis zueinander ein, ihre
40
(30) Das Konzert müsste in einer Stunde zu Ende sein. „The concert should end in an hour.‟
Equivalent to negated MÜSSEN is negated BRAUCHEN, which Zifonun et al.
(1997:1252) classify as belonging to the periphery of modal verbs. In formal speech,
it usually governs an infinitive with zu, but it can also govern an infinitive without zu
(esp. in informal speech):
(31) Du musst heute nicht in die Schule gehen. = Du brauchst heute nicht in die Schule (zu) gehen. „You don‟t have to go to school today.‟
2.3.1.2.2 KÖNNEN
Contrary to MÜSSEN, KÖNNEN implies possibility rather than necessity. This can
either be motivated positively, i.e. that there are potential conditions that allow an
implementation or an assumption or negatively, i.e. that there are no known obstacles
that prevent the proposition. Conversely, negated KÖNNEN implies that there are
circumstances that prevent the proposition. Conditions that allow a potential
implementation can be external circumstances or necessary means being available
or simply that there are no known obstacles:
(32) Da ihre Eltern die Studiengebühren bezahlen, kann die junge Frau die Universität besuchen. „Because her parents pay the tuition fees, the young woman can go to university.‟
but they can also be internal circumstances where KÖNNEN denotes abilities,
qualities or habits of human beings30:
(33) Der kleine Junge kann sehr gut schwimmen. „The little boy can swim very well.‟
A third variant of non-epistemic KÖNNEN applies when it is used in the sense of
giving permission as an alternative to DÜRFEN:
(34) In einigen Ländern kann man ohne Waffenschein eine Pistole tragen. „In some countries you can carry a pistol without a gun licence.‟
If KÖNNEN is used, the speaker indicates that there are no obstacles that would
prevent an implementation (in example (34) there is no law against it); if DÜRFEN is
soziale Rolle, ihre Erfahrungen, Erinnerungen und Erwartungen, natürlich auch ihr „Vorwissen“, alles woran sie denken.“ (Brinkmann 1971:730)). During the production of a text (in the broadest sense including e.g. oral conversation), the horizon of speaker/writer and hearer/reader is typically unconsciously taken for granted. In the text itself the horizon of the speaker/writer leads to presuppositions, i.e. contents that are implied rather than explicitly uttered and are taken as given. The notion of Horizont will play an important role in the discussion of the subjunctive mood (see 2.3.5 below).
30 Cf. “dynamic modality” (section 2.1.3).
41
used, the focus is on an external party giving their consent for an implementation, but
in everyday language use this distinction is often neglected.
In epistemic sense, KÖNNEN implies that indications exist that a certain
assumption is possible without, however, categorically excluding other possibilities
(as in the case of MÜSSEN):
(35) Ein heimlicher Verehrer kann die Blumen geschickt haben. „A secret admirer may have sent the flowers.‟
As with MÜSSEN, the Konjunktiv II form of KÖNNEN – KÖNNTE – shifts the
proposition beyond the given horizon, adding a notion of tentativeness to the
utterance without changing the basic value of the modal verb.
2.3.1.2.3 WOLLEN
WOLLEN presents an utterance in a way that the execution of the given proposition
is dependent on the subject alone; other facets are neglected. In the realm of
implementations, this direction from the subject can be motivated by the subject‟s
willingness, intention, plan or decision. In announcements WOLLEN can evoke
expectations: The speaker decides on the direction they want to take:
(36) Im folgenden Bericht will ich die Gründe für meine Vorgehensweise erklären. „In the following report I want to explain the reasons for my course of action.‟
In this respect, WOLLEN expresses the transition from willingness to execution and is
therefore close to WERDEN (see below). This becomes evident when we look at the
English translation of (36), where „…I will explain…‟ would also be an adequate
translation.
It is noteworthy that the direction for implementation in the sense of „intention‟ can
also be ascribed to inanimate objects. In these instances, WOLLEN is often negated
and thus expresses an unfulfilled expectation:
(37) Das Feuer will einfach nicht ausgehen. „The fire just won‟t go out.‟
In the realm of information, the notion of direction pertains to the question who is
responsible for the information conveyed in the utterance (evidential). In the case of
WOLLEN, the direction originates from the subject; this means that the subject is
portrayed as being responsible for the information, therefore such an utterance can
be paraphrased as „The subject claims…‟:
42
(38) Sie will den Axtmörder gesehen haben. = Sie behauptet, den Axtmörder gesehen zu haben. „She claims to have seen the axe murderer.‟
2.3.1.2.4 SOLLEN
SOLLEN assumes the opposite direction to WOLLEN. The grammatical subject
receives its direction from an external authority, which usually remains unspecified
and has to be deduced from context (in this respect, SOLLEN is similar to the
passive). If the direction that the external authority gives pertains to an
implementation, we are typically dealing with instructions or commands:
(39) Die Reporter sollen sich aus dem Krisengebiet zurückziehen. ‟The reporters are told to withdraw from the conflict area.‟
Depending on the nature of the authority and the person of the subject, expectations,
promises, commandments or moral obligations or prohibitions can all be expressed in
utterances with the modal verb SOLLEN. If the Konjunktiv II (SOLLTE) is used, the
meaning changes from instruction to suggestion, the execution of which remains
uncertain, as the subjunctive causes a shift beyond the given horizon. The external
authority in this case is not an agent but unknown factors that the speaker has no
control over:
(40) Die Regierung sollte die Steuern erhöhen. ‟The government should raise taxes.‟
SOLLTE (Konjunktiv II) – but not SOLLEN – can also be used in order to mark a
piece of information as an assumption. In this case the two functions combine in the
same way as for implementations, so that the information is marked as beyond the
given horizon and dependent on unknown factors:
(41) Der Brief sollte mittlerweile angekommen sein. „The letter should have arrived by now.‟
On top of that, SOLLEN can be used analogous to WOLLEN as an evidential. As
SOLLEN fundamentally implies an external agent, in its use as evidential this external
agent is made responsible for the information conveyed in the utterance. It is neither
the speaker nor the grammatical subject that makes a claim but rather an (often)
unspecified „someone‟ or „everyone‟ – hearsay:
43
(42) Er soll sich eine Luxusvilla in Monaco gekauft haben. „He is said to have bought a luxury villa in Monaco.‟
2.3.1.2.5 DÜRFEN
The modal verb DÜRFEN implies two directions. The grammatical subject has one
direction (like WOLLEN), but this alone is not enough for an implementation. What is
required is a second direction from an authoritative instance that legitimises this
direction. It is, however, different from SOLLEN in that with DÜRFEN the grammatical
subject is not subjected to the direction of an external agent – it has its own direction,
but accepts the authority of the external agent.
As always, this fundamental relation can be applied in both types of modality.
Concerning implementations there are two variations (both realised by the same
form), depending on whether the external instance is a person or an inanimate
institution. A person gives permission (43), an inanimate institution legitimises, e.g. by
law (44):
(43) Ich darf heute abend auf die Party gehen. „I‟m allowed to go to the party tonight.‟
(44) Man darf 200 Zigaretten zollfrei einfühen. „You are allowed to import 200 cigarettes duty free.‟
As with SOLLTE, DÜRFEN can only be used for assessing the validity of
information in the Konjunktiv II form DÜRFTE. The subjunctive form shifts the
legitimisation that DÜRFTE infers onto an assumption that the speakers deems
feasible31 and for which they expect agreement from the external authority:
(45) Das dürfte Schwierigkeiten geben. „That will probably cause problems.‟
2.3.1.2.6 MÖGEN
MÖGEN combines two fundamental elements: on the one hand, it expresses a
direction for implementation and on the other hand it implies that there are no known
obstacles that oppose the direction. With regards to implementations this use of
MÖGEN is largely confined to very formal or literary language. The speaker indicates
that the subject is free to implement the given proposition while at the same time
acknowledging that the subject might want to follow a different direction:
31
This assessment is corroborated by Mortelmans (1997:211), who ascertains that DÜRFTE is the most grammaticalised of the epistemic modal verbs and therefore carries the highest degree of „speaker orientation‟, i.e. based on the speaker‟s subjective belief state/attitude towards the proposition.
44
(46) Gegen gesetzliche Sonderregelungen oder besondere Schutzvorschrifen für „Ehen ohne Trauschein“ hat sich Justizminister Engelhard (FDP) in Bonn ausgesprochen: „Wer rechtliche Regelungen haben will, der mag heiraten“, betonte er.32 „Justice Secretary Engelhard (FDP) has spoken out against special laws or protective regulations for domestic partnerships: “Those who want legal regulations might like to get married” he pointed out in Bonn.‟
In non-literary texts, MÖGEN is mostly used in a specific variant, namely to
express liking or affection or, more often, dislike (negated MÖGEN), albeit in a way
that acknowledges the existence of other directions (and is therefore less forceful
than WOLLEN):
(47) Ich mag heute nicht schwimmen gehen. „I don‟t want to go swimming today.‟
If combined with an accusative complement rather than an infinite verb, MÖGEN
expresses a general liking/affection or dislike that is independent of a particular time
or situation:
(48) Ich mag keine Schokolade. „I don‟t like chocolate.‟
As mentioned before, the Konjunktiv II form MÖCHTE has almost developed into a
modal verb in its own right, that is close in meaning to WOLLEN but less forceful due
to the usual subjunctive shift beyond the given horizon. It is this characteristic that
allows MÖCHTE to be used to express wishes in an unobtrusive way, which is why it
often co-occurs with other markers of wishes such as lieber (rather) or gerne :
(49) Ich möchte gerne wissen, was er denkt. „I would quite like to know what he‟s thinking.‟
When applied to the realm of information, the speaker can use MÖGEN in order to
express that they do not know of any indications that opposes an assumption. In this
sense, it is close in meaning to epistemic KÖNNEN. The difference is that MÖGEN
inherently indicates that other directions are not excluded and therefore MÖGEN is
usually used in concessive contexts:
(50) Das mag zwar arrogant klingen, aber es ist die Wahrheit. „This may sound arrogant, but it is the truth.‟
MÖCHTE is not available in epistemic modality.
32
From: Mannheimer Morgen, 1./2.8.1987 (attested in Zifonun et al. 1997:1895).
45
2.3.1.2.7 WERDEN
WERDEN occupies a special position in the system of modal verbs. It has
morphological and syntactic properties that set it apart from the other modal verbs,
yet semantically it has some decisive properties in common with them that warrant its
inclusion in the modal verb system.
There are three distinct syntactic types of WERDEN that also have different
meaning33:
WERDEN as main verb:
has the past participle form geworden
has a full paradigm of verb forms
can combine with modal verbs
combines with adjectives or noun phrases in the sense of to become/to get
(51) Die Studenten werden vor den Prüfungungen immer sehr nervös. „The students always get very nervous before the exams.‟
WERDEN as auxiliary verb:
is used to form the passive
has the past participle form worden
has a full paradigm of verb forms except the imperative
can combine with modal verbs
(52) Das Bild wurde gestern gestohlen. „The painting was stolen yesterday.‟
WERDEN as (semi-)modal verb:
like the other modal verbs, WERDEN in this sense combines with a verb in the
infinitive without zu (bare infinitive)
unlike the other modal verbs, however, it
has no past participle
has a restricted verb paradigm (no imperfect and no imperative)
cannot combine with modal verbs
can only have epistemic meaning (not deontic)
(53) Es hat geläutet. – Das wird der Briefträger sein. The doorbell has rung. – That will be the postman.
The first two types do not carry modal meaning, so it is only the third type of
WERDEN that is of interest here. There are, however, some additional complications
33
Cf. Engel 1988:468ff.
46
that need to be explained. In most German grammars (including learner grammars,
e.g. Helbig and Buscha 2001) WERDEN + bare infinitive is traditionally dealt with
under the heading „future tense‟. An example would be
(54) Die Arbeiter werden morgen streiken. „The workers will go on strike tomorrow.‟
While this is true in parts, there is now general consensus that this tells only half the
story. We can clearly see from example (53) above that this construction can also
refer to present time rather than future time. The present tense, on the other hand,
can also be used to refer to future time:
(55) Die Arbeiter streiken morgen. ?„The workers go on strike tomorrow.‟
The temporal adverbial morgen clearly identifies the future reference; without it the
sentence could refer to either present time or future time. Zifonun et al. (1997:1900),
Helbig and Buscha (2001:137), Nehls (1986:120) and Itayama (1993:233) therefore
all agree that, in terms of temporal reference, present tense and WERDEN + infinitive
constructions can coincide. Therefore, when used with present or past time
reference, rather than a temporal characteristic, WERDEN adds a modal
characteristic to the sentence. It expresses a confident assumption which, according
to Brinkmann (1971:398), is based on the speaker‟s experience, where the level of
confidence in the fact that the proposition is true is not as high as if it was expressed
in the present tense. If we want to make a sentence in the present tense equivalent to
the same sentence with WERDEN (examples (56) and (57)), we therefore need to
add some kind of lexical modal marker:
(56) Er wird im Büro sein. „He‟ll be in the office.‟
(57) Er ist wahrscheinlich/wohl im Büro. ‟He‟s probably in the office.‟
However, WERDEN can, of course, also refer to future time (see example (54)). In
these cases, there is a debate whether all of them carry modal meanings or not.
Zifonun et al. (1997:1901) and Helbig and Buscha (2001:138) state that “these types
can contain a factor of modality (of an assumption), but they do not necessarily do
so.”34 Both do not, however, offer any explanation on how to determine whether a
sentence has an epistemic „hue‟ or not. Itayama (1997:234) distinguishes different
nuances in WERDEN constructions with future reference, but generally assumes a
34
“Diese Variante kann einen Modalfaktor (der Vermutung) enthalten, muss es aber nicht. ”
47
modal element for all of them. He identifies person of the subject, verb semantics and
role of the speaker as factors that influence to what degree an utterance entails
modal meaning. For instance, a sentence like (58), with a first-person subject, a verb
that expresses a volitional action and a speaker who has direct influence on the
realisation of the action, only carries a marginal epistemic modal element and can be
labelled with the additional tag „intention‟.
(58) Ich werde den Entwurf heute noch abschicken. „I will send the draft off today.‟
In a sentence like (59), on the other hand, with a third-person subject, a non-
volitional verb and a speaker that has no influence on the realisation of the
proposition, the modal element is much more prominent.
(59) Die armen Länder werden nicht von diesen Spenden profitieren. „The poor countries will not profit from these donations.‟
On top of that, WERDEN seems to be quite flexible with regards to the degree of
certainty. It can co-occur with modal adverbials like vielleicht, wahrscheinlich
(perhaps) etc. but also with adverbials like sicher, bestimmt, gewiss (certainly,
definitely). Zifonun et al. (1997:1902) therefore place it between the classic opposition
„epistemically necessary‟ (MÜSSEN) and „epistemically possible‟ (KÖNNEN). In the
same vein, Brünner and Redder (1983:83) report that in a test they carried out to
determine the degree of certainty that epistemic modal verbs express, WERDEN was
ranked between MÜSSTE and DÜRFTE, which seems to corroborate Zifonun et al.‟s
analysis.
In summary, we can therefore say that although WERDEN exhibits some morpho-
syntactic differences to the other modal verbs, semantically it behaves like a modal
verb and will therefore be included in the present study together with MÜSSEN,
KÖNNEN, SOLLEN, WOLLEN, DÜRFEN and MÖGEN.
2.3.2 Modal adverbials
We have seen in the previous sections that the concept of modality itself and the
category of modal verbs are saturated with a wide range of disparate terminology
based on different theoretical frameworks. This also applies to the category described
in this section and it is therefore necessary to first delineate the terminology that will
be used in this study. The terms Modalwörter – „modal words‟ (Admoni 1982; Helbig
and Helbig 1990), Modaladverben – „modal adverbs‟ (Brinkmann 1971) and
48
Modalpartikeln35 – „modal particles‟ (Zifonun et al. 1997; Engel 1988) have all been
used, each denoting different sets of items in the respective treatises.
Brinkmann (1971) only considers modal adverbs in this category and since his
account is unsatisfactorily brief for our purposes I will only introduce a short summary
of his ideas. I will then turn to Zifonun et al. (1997) who offer a definition of modal
adverbials that is in accord with Brinkmann‟s but provide a much more
comprehensive account of both syntactic and semantic characteristics of modal
adverbials.
2.3.2.1 Brinkmann‟s (1971) account of modal adverbs
Brinkmann (1971:400) defines modal adverbs as “a class of fixed lexical units that do
not modify the content of the verb but the conception of the whole sentence. We
could also call them „sentence adverbs‟”.36
They are only available for the modification of the validity of information (epistemic
modality) as they denote a particular aspect under which the speaker evaluates the
information. Brinkmann identifies three functions of this modification:
Asserting or negating/restricting the validity of information:
e.g. sicher, gewiss, in jedem Fall, zweifellos (surely, certainly, in any case,
doubtlessly)
keinesfalls, unter keinen Umständen/kaum, schwerlich (not at all, under no
circumstances/hardly)
Marking the content of a proposition as fact or assumption:
e.g. tatsächlich, wirklich, natürlich (as a matter of fact, really, of course)
offenbar, wahrscheinlich, vermutlich, vielleicht, möglicherweise (apparently,
probably, presumably, perhaps, maybe)
Assessing a proposition emotionally:
e.g. leider, hoffentlich, glücklicherweise (unfortunately, hopefully, fortunately)
Brinkmann adds briefly that modifications with modal adverbs can often be
transformed into clauses in the following manner:
35
The term Modalpartikeln is particularly tricky as some linguists (e.g. Weinrich 2005:841) – incorrectly – use it to refer to a group of items like ja, doch, eben etc. that is better subsumed under the term Abtönungspartikel (usually referred to as „illocutive particles‟ or „discourse particles‟ in English). These Abtönungspartikel have no modifying effect.
36 “Als Modaladverb wird eine Klasse unveränderlicher sprachlicher Einheiten verstanden, die nicht den Ihnhalt des Verbums, sondern die Auffassung des Satzes modifizieren […]; man könnte auch von „Satzadverb“ sprechen.”
49
(60) Sie ist wahrscheinlich schon nach Hause gegangen. „She has probably already gone home.‟ = Es ist wahrscheinlich, dass sie schon nach Hause gegangen ist. „It is probable, that she has already gone home.‟
This means that he does acknowledge the fact that this group of modal expressions
is not restricted to the part of speech of adverbs (as his category term „modal
adverbs‟ suggests), but does not address this issue in any more detail.
2.3.2.2 Zifonun et al.‟s (1997) account of modal adverbials
Unlike Brinkmann, Zifnonun et al. (1997:861ff) approach the category from a
functional point of view under the heading Modalfunktion (modal function). Modal
functions transform categorical assertions into evaluative statements, i.e. they modify
the validity of a proposition in that they either express the speaker‟s evaluation of the
existence/occurrence of a certain situation or the speaker‟s assessment of the
probability with which a certain situation exists or will occur. This is, in essence, the
basic definition of epistemic modality, and thus congruent to the function of modal
verbs: “The use of modal functions is not the only way in which we can bring to
account that we do not possess all the information to make categorical statements. A
comparable effect can be achieved with a modalised predicate – actualised by a
modal verb construction.”37 (Zifonun et al. 1997:862)
Modal functions modify sentence meanings. They are therefore syntactically
realised by sentence adverbials. Unlike other adverbials, they do not qualify the
predicate of the utterance but operate on the whole sentence in such a way that they
express the speaker‟s attitude towards the proposition. Zifonun et al. (1997:1126ff)
deal with this category under the term Modale Satzadverbialia (modal sentence
adverbials), in order to distinguish them from other types of sentence adverbials. In
the following, I will refer to this category simply as modal adverbials. Zifonun et al.
(1997:1126) observe that adjectives (61), modal particles (62) prepositional phrases
(63) and subordinate clauses (64) can all fulfill the syntactic function of modal
adverbials:
37
“Die Verwendung von Modalfunktionen ist nicht die einzige Form, in der wir in Rechnung stellen können, daß wir unvollständig informiert sind. Eine vergleichbare Wirkung kann durch ein modalisiertes Prädikat – realisiert als Modalverbgefüge – erreicht werden.“
50
(61) Das Geld wurde offensichtlich gestohlen. ‟The money was obviously stolen.‟
(62) Vielleicht hat es ein Angestellter unterschlagen. ‟Maybe an employee embezzled it.‟
(63) Zum Glück ist die Firma versichert. ‟Luckily the company is insured.‟
(64) Wie zu vermuten ist, wird es eine interne Untersuchung geben. ‟It is assumed that there will be an internal inquiry.‟
This syntactic category of modal adverbials is then divided into subcategories
according to semantic criteria which are based on the following method of conclusion:
There are three possibilities for the effect that these adverbials can have on the
meaning of a sentence. These effects can be summarised in the following logical
conclusions where a sentence s‟ is derived from sentence s by dropping the sentence
adverbial: Either from s follows s‟, or from s follows „possibly (s‟)‟, or from s follows
„not (s‟). All adverbials are assigned to one of these three categories, depending on
what effect they achieve with respect to the relation between sentence s (with
adverbial) and s‟ (without adverbial). The three categories of modal adverbials are
labeled „assertive adverbials‟, „modally qualifying adverbials‟ and „negative
adverbials‟:
Assertive adverbials:
From s follows s‟.
For instance, from sentence
s: Er hat das Geld wirklich gestohlen („He has actually stolen the money‟) follows
s‟: Er hat das Geld gestohlen. (‟He has stolen the money.‟)
This group of adverbials can then be further divided into:
o Purely assertive adverbials:
They comprise the relatively small group of adjectives tatsächlich, wirklich
(actually) and the modal particles gewiss/gewisslich (certainly). They are called
„purely assertive‟ as they do not convey any additional information on the rest of
the sentence other than the emphasised assertion of its validity.
o Evaluative assertive adverbials:
On top of the assertion of the validity of an utterance these adverbials express
some form of evaluation by the speaker. This can be positive, e.g. through
modal particles like glücklicherweise (luckily), Gott sei Dank (thank God) etc;
the evaluation can be reservedly positive with expressions like immerhin,
51
wenigstens, zumindest (at least) or negative with modal particles like leider
(unfortunately), bedauerlicherweise (regrettably) etc.
Other modal particles in this group express evaluations that cannot be arranged
on a positive – negative scale, but add individual evaluative dimensions. They
are all formed from adjectives (which signals the respective evaluative
dimension) by adding the filler-infix -er- and the suffix –weise, so that e.g.
verständlicherweise (understandably) adds the dimension of empathic
reasoning, bemerkenswerterweise (remarkably) adds the dimension of
relevance.
o Evidence-focused assertive adverbials
The elements in this group express, on top of the assertion, a reference to inter-
subjective knowledge and therefore add an extra focus on evidence to the
assertion. These are adjectives like offensichtlich (evidently), nachweislich
(demonstrably) and modal particles like bekanntlich (as is generally known) and
zweifellos (undoubtedly). In addition to this, this group entails adverbials that
express the strongest focus on evidence – necessity. These are modal particles
like notwendigerweise (necessarily), adjectives like notwendig (necessary),
unausweichlich (inescapably) etc. and prepositional phrases like mit
Notwendigkeit (with necessity). It has to be pointed out here that in Zifonun et
al.‟s (1997:1131) analysis, these expressions do not just pertain to epistemic
modality, but can also apply to non-epistemic contexts, e.g. in the sense of
normative obligations. This is important as it gives justification for the inclusion
of lexemes other than modal verbs also for the expression of non-epistemic
modal meanings.
Modally qualifying adverbials
From s follows „possibly (s‟)‟.
For example, from sentence s:
Wahrscheinlich hat er das Geld gestohlen. („He has probably stolen the money.‟)
follows neither Er hat das Geld gestohlen. („He has stolen the money.‟)
nor Er hat das Geld nicht gestohlen. („He has not stolen the money.‟)
but s‟: Möglicherweise hat er das Geld gestohlen. („It is possible that he has stolen
the money.‟)
Adverbials in this group comprise adjectives like angeblich (allegedly),
anscheinend (apparently), wahrscheinlich (probably) etc, modal particles like
bestimmt (surely), vielleicht (maybe), möglicherweise (possibly) etc and formulaic
short forms like wie angenommen (as assumed). It is evident here that these
modal adverbials can be matched up with epistemic modal verb meanings. Modal
52
adverbials like sicher (certainly), vermutlich (presumably) and vielleicht (maybe)
express the scale of a speaker‟s assessment of the validity of a proposition in
roughly the same way as the epistemic modal verb meanings MÜSSEN –
DÜRFTE – KÖNNEN. Adverbials like angeblich (allegedly) or wie verlautet
(reportedly) indicate that the speaker does not take responsibility for the
information conveyed in the utterance in the same way as evidential SOLLEN.
Negative adverbials
From s follows „not (s‟)‟.
From s: Er hat sich auf keineswegs auf die Ferien gefreut. („He did not, by any
means, look forward to the holidays.‟) follows
s‟: Er hat sich nicht auf die Ferien gefreut. („He did not look forward to the
holidays.‟)
This relatively small group of negative adverbials entails the negation particle nicht
(not). This is the least specific negative adverbial, as it simply expresses negation,
whereas other adverbials in this group imply additional meaning components on
top of the negation. Adverbs like nie/niemals (never) or nirgends/nirgendwo
(nowhere) and prepositional phrases like zu keiner Zeit (at no time) additionally
specify the negation within a certain temporal or spatial scope. The expanded
negative particle gar nicht/überhaupt nicht (not at all), modal particles like
keinesfalls, keineswegs, mitnichten (by no means) and prepositional phrases like
auf keinen Fall (in no case) intensify the negation and often negate a proposition
that is to be expected or assumed.
The modal particles fast and beinahe (nearly) take up a special position here. On
top of the negation they express that the occurrence of the event was imminent or
very likely. We can therefore conclude from „fast (s)‟ a past „wahrscheinlich (s)‟
which gives fast and beinahe a secondary modally qualifying meaning. This is why
these modal particles often occur with the Konjunktiv II to indicate the counter-
factuality of the proposition:
(65) Wegen des schlechten Wetters wäre er fast abgestürzt. Due to the bad weather was-KONJII he nearly crashed ‟Due to the bad weather he nearly crashed.‟
It has to be noted here that other descriptions of modal adverbials, e.g. Helbig and
Buscha (2001:436) exclude all of these negation words from the category of modal
expressions for syntactic and semantic reasons. Their explanation is convincing
for the negation particle nicht (not) – a proposition that is negated by nicht does
not contain any modal factor in the same way that I argue that an affirmative
proposition in the indicative without any explicit lexical or grammatical marker of
53
modality is unmarked in terms of modality. The other intensifying adverbials in this
group, like keineswegs or auf keinen Fall, do, however, add additional modally
qualifying meaning to the sentence (analogue to the purely assertive adverbials)
and will therefore be included in the analysis.
Here ends Zifonun et al.‟s (1997) description of modal adverbials. We have seen
that there is an abundance of alternative adverbial expressions to modal verbs that
can serve as modal functions. While modal verbs offer a wide semantic range (and
the possibility for ambiguity) due to their polysemic nature, modal adverbials are more
specific in the type of modality they denote and also in additional meaning overtones
that they can carry. It has to be pointed out, however, that this characterisation is not
entirely universal. Hofmann (1979:3f) calls attention to the fact that, for example,
modal adjectives like possible and necessary can be used epistemically and non-
epistemically – this has already been mentioned briefly in the section on evidence-
focused adverbials above, where Zifonun et al. (1997:1131) make a similar
observation for German. The adjective möglich (possible), for instance, can be used
to express possibility with regard to the validity of information (66) or with regard to an
implementation (67):
(66) Es ist möglich, dass er das Geld gestohlen hat. „It is possible that he has stolen the money.‟
(67) Es war ihm möglich, das Geld rechtzeitig zu verstecken. „It was possible for him to hide the money in time.‟
This raises another important issue. Zifonun et al. (1997) focus their description of
modal adverbials almost exclusively on epistemic modality – Brinkmann (1971)
explicitly states that modal adverbs are only available for epistemic modality. But if,
as can be seen from the example above, some modal adverbials can also be used as
alternatives to non-epistemic modal verbs, then it must be considered which other
expressions can fulfil the same semantic functions as non-epistemic modal verb
meanings and these expressions ought to be included in the analysis. Gutknecht and
Rölle (1988:165f) address this point from a translational perspective:
“Within translational contexts the inclusion of modal expressions other than the
modal verbs is determined by the fact that the modal verb systems in German
and English are not formally congruent (isomorphic) […] This is evident, for
example, in the fact that WOLLEN is usually not expressed by the modal verb
WILL in English, but by a main verb like WANT or INTEND. This raises the
54
question whether verbs like WANT should be marked as modal […] and Perkins
(1983:98) actually categorises WANT NP TO as a modal main verb.”38
If we then take this argument another step further and translate back into German,
then it becomes obvious that e.g. the main verb beabsichtigen (intend to) ought to be
included in an analysis of modal expressions, because it serves as an alternative to
WOLLEN in the meaning of „intend‟. This brings out two crucial points:
Firstly, modal expressions other than modal verbs are not only available for
epistemic modality, but also for non-epistemic modality. Unlike modal verbs, which
can be ambiguous between the different types of modality, however, these
expressions are, for the most part, different from the ones used for epistemic senses.
Secondly, not only adverbial constructions can fulfil modal functions but also
lexical verbs. This is discussed in more detail in the next section.
2.3.3 Modal lexical verbs and quasi-modal verbs
As can be seen from the examples in the previous section, there are a number of
different syntactic ways of imposing additional modal meaning onto a proposition.
Often, these different syntactical solutions can be brought about by paraphrasing the
sentence using the same lexeme as a different part of speech, e.g. from adverb to
matrix clause with adjective:
(68) Möglicherweise hat er das Geld gestohlen. „He has possibly stolen the money.‟
= Es ist möglich, dass er das Geld gestohlen hat. „It is possible that he has stolen the money.‟
These transformations can not only involve adjectives, modal particles,
prepositional/nominal constructions or adverbs, as Zifonun et al. (1997) suggest, but
also lexical verbs, as the following examples show:
38
“Die Einbeziehung anderer Modalausdrücke als der der MVen selbst ist im Rahmen translatorischer Fragestellungen durch die Tatsache bedingt, daß die MV-Systeme in D und E formal nicht deckungsgleich (isomorph) sind […] Dies zeigt sich z.B. daran, daß WOLLEN in E im allgemeinen nicht mit dem MV WILL wiedergegeben wird, sondern mit einem Vollverb wie WANT oder INTEND. Damit ergibt sich die Frage, ob auch Verben wie WANT als modal zu kennzeichnen sind […] und Perkins (1983:98) bezeichnet denn auch WANT NP TO als modales Vollverb.”
55
(69) Hoffentlich ist er nicht ernsthaft krank. „Hopefully he is not seriously ill.‟
= Wir hoffen, dass er nicht ernsthaft krank ist. „We hope that he is not seriously ill.‟
(70) Vermutlich hat er nur eine Erkältung. „He has probably only got a cold.‟
= Ich vermute, dass er nur eine Erkältung hat. „I assume that he has only got a cold.‟
As lexical verbs like hoffen (hope), vermuten (assume), glauben (believe), denken
(think), scheinen (seem) etc. express similar kinds of modal meaning as their
adverbial counterparts and as they can, in part, compete semantically with the modal
verbs (cf. also Diewald 1999:175), these verbs are included in the analysis under the
term „modal lexical verbs‟ (cf. Perkins 1983:94).
In German, some of the lexical verbs that convey modal meaning exhibit not only
semantic but also a certain degree of syntactic affinity to the modal verbs in the fact
that they combine with an infinitive (although always with the particle zu). This is why
the term quasi-modale Verben (quasi-modal verbs) has been suggested for this
group (Weinrich 2005:315). Engel (1988:477f) calls them Modalitätsverben (modality
verbs) and defines them as lexical verbs that combine with an infinitive and
semantically compete with modal verbs, although they are usually more specific in
their meaning:
(71) Sie versteht es, ihre Kollegen zu motivieren. „She knows how to motivate her colleagues.‟
(72) Sie kann ihre Kollegen motivieren. „She can motivate her colleagues.‟
(71) can be paraphrased using the modal verb KÖNNEN (72), but the lexical verb
expresses more specifically and unambiguously the meaning of „having the ability to‟
than the modal verb.
The items that are included in this category differ, again, from one description to
the next. Many of the verbs that Engel (1988:477) lists are nowadays dated or
restricted to formal or literary language, e.g. anheben zu (be about to), belieben zu
(choose to), gedenken zu (intend to), wissen zu (know how to). Weinrich (2005:315)
includes in his list some additional verbs and participial constructions that are more
commonly used, such as imstande sein zu (be able to), planen zu (plan to),
beabsichtigen zu (intend to), gezwungen sein zu (be forced to), verpflichtet sein zu
(be obligated to).
56
Zifonun et al. (1997) present a finer grained syntactical analysis in which three
verbs are analysed as Halbmodale (half-modals): drohen zu (threaten to), scheinen
zu (seem to), pflegen zu (usually do). This analysis is based on syntactical criteria,
however, so that immediately the caveat is made that only scheinen (seem) is
semantically classed as modal. The other verbs that Engel (1988) identifies as quasi-
modal verbs are analysed as verbs with verb complement, but they are not discussed
in any more detail in Zifonun et al.‟s (1997) account. As this study follows semantic
criteria for the identification of modal expressions, all quasi-modal verbs and modal
lexical verbs are included in the analysis. In the analysis of the learner data, both
types of verbs are grouped together with modal adverbials under „periphrastic modal
expressions‟.
Both Zifonun et al. (1997) and Brinkmann (1971), however, present two of the
verbs separately that the other grammars include in the group of quasi-modal verbs:
HABEN ZU and SEIN ZU. I will follow their classification and deal with them in the
next section.
2.3.4 Modal infinitives HABEN ZU and SEIN ZU
In his account of modality, Brinkmann (1971:361) only deals with HABEN ZU (have
to) and SEIN ZU (be to) as verbal alternatives to modal verbs. The term for these
constructions is modaler Infinitiv (modal infinitive) (cf. Brinkmann 1971; Gelhaus
1977; Langhoff 1980). Zifonun et al. (1997) analyse these two forms as the
immediate periphery of modal verbs. Modal infinitive constructions are less common
than modal verbs and adverbials, but they do present an alternative to expressing
deontic modal meanings. The unique quality of modal infinitives is that either of them
can express both necessity and possibility, i.e. they can compete with the modal
verbs MÜSSEN and KÖNNEN; SEIN ZU can additionally be used in the meaning of
obligation or advice (SOLLEN/SOLLTE) or even permission (DÜRFEN).
Helbig and Buscha (2001:112) explain that HABEN ZU is used more often to
express deontic necessity (example (73)) than possibility (74). However, the
possibility reading is more likely in negated sentences. In the end only the context
can clarify which is intended in a particular instance.
57
(73) Ich habe mit dir zu reden. I have with you to talk „I have to talk to you.‟
(74) Was hast du zu berichten? What have you to tell „What can you tell me?‟
From a contrastive point of view, Nehls (1986:35) adds that both German HABEN
ZU and English have to for deontic necessity are only used in contexts where the
source of the necessity is unimportant. However, have to has a much wider range of
application than HABEN ZU, so that all instances of German HABEN ZU (where
denoting necessity) can be translated as English have to but not vice versa.
HABEN ZU constructions have active meaning where the agent subject is typically
a human being. They typically combine with verbs that denote actions, processes or
states that can be influenced or initiated by human beings.
SEIN ZU constructions, on the other hand, have passive meaning where the agent
is usually unnamed. They also combine with physical action verbs, but also often with
verbs of perception, saying/thinking verbs and verbs expressing subjective attitude.
With SEIN ZU the ambiguity between necessity and possibility is even more
pronounced:
(75) Das Fenster ist zu öffnen. The window is to open 'The window must be opened/The window can be opened.'
Again, only the context will resolve the ambiguity. This often happens through further
modifications in the sentence39:
(76) Das Fenster ist jeden Tag für 1 Stunde zu öffnen. The window is every day for 1 hour to open „The window is to be opened for 1 hour every day.‟
(77) Das Fenster ist leicht zu öffnen. The window is easy to open „The window can easily be opened.‟
Helbig and Buscha (2001:112) note that converse to HABEN ZU, sentences with
SEIN ZU express a possibility more often than a necessity (which is typically
restricted to rules and regulations).
There are, however, instances where the variability of HABEN ZU and SEIN ZU
with regard to their modality is deliberately exploited. Langhoff (1980:453) sees it as
“characteristic” for modal infinitive constructions “that they are not tied down to only
39
For a detailed discussion of these constructions as instances of modal infinitives see Holl (2001:219).
58
one perspective, but that they can communicate in a linguistically condensed way the
complex attitude that a speaker often takes towards complicated issues.”40 He offers
as an example the following sentence (Langhoff 1980:269), in which the speaker
avoids specifying whether they want to express a possibility, an obligation, a
necessity or even a permission:
(78) Psychologen, Psychiater und Sexualwissenschaftler streiten sich darum, ob der Baron für die Bluttat verantwortlich zu machen ist. = … verantwortlich gemacht werden kann/soll/muss/darf. „Psychologist, psychiatrists and experts in sexology argue whether the baron can/should/must/may be blamed for the massacre.‟
This concludes the synopsis of lexical means of expressing modality. What remain
to be discussed are the modal qualities of the subjunctive mood in German.
Before that, however, one more important issue is to be addressed with regard to the
large and diverse groups that is referred to here collectively as „periphrastic modal
expressions‟41 (modal adverbials and modal lexical verbs/quasi-modal verbs) and the
modal infinitives. So far, their „modality‟ has been demonstrated by using them as
paraphrases of modal verb meanings, effectively explaining their inclusion in the
investigation at hand by highlighting the semantic similarity between them and modal
verbs. While this is certainly justified, it is at the same time essential to be aware that
periphrastic modal expressions are not synonymous with modal verbs. Perkins
(1983) makes this point very accurately as one of the fundamental principles
underlying his approach to modality. He argues that discussions of modality in
linguistics have been preoccupied with modal verbs because these constitute the
only formally coherent class of modal expressions, while other modal expressions
have been mentioned only in so far as they serve as paraphrases to illuminate the
meaning of modal verbs. These expressions are usually declared semantically
equivalent to modal verb meanings, the only difference being one of stylistic
characterisation. Perkins (1983:19) addresses this issue:
40
“[Für die modalen Infinitivkonstruktionen ist es] charakteristisch, daß sie nicht auf genau eine Sehweise fixiert sind, sondern in sprachlich kondensierter Form jene umfassende komplexe Haltung kommunizieren können, die ein Sprecher ja häufig vielschichtigen Sachverhalten gegenüber einnimmt.”
41 N.B. The term „periphrastic modal expressions‟ suggests that the notion of modality is expressed in more than one lexical/grammatical item within the utterance. This encapsulates the notion that these expressions can be multi-word units, such as „meiner Meinung nach‟ (in my opinion) or even clauses. In this study, however, it also includes single-word lexical items, such as the adverb „vielleicht‟ (perhaps), even though these are, strictly speaking, not „periphrastic‟.
59
“However, the relative „appropriateness‟ of different modal expressions, even
on the rare occasions when it is actually considered, is never dealt with in any
detail, and more often it is ignored altogether […] I shall argue later that in fact it
is the case that the way in which modal auxiliary verbs are used to express
modality is different in some sense from that in which non-auxiliary modal
expressions are used.”
Perkins‟ explorations obviously pertain exclusively to the English language, but the
factors that he shows to be relevant for the distinction between modal verbs and
periphrastic modal expressions are certainly worth considering for the German
language, too. As mentioned before, Perkins (1983:100) postulates that modal verbs
are the least formally explicit of all modal expressions. They only specify the nature of
the relationship (e.g. necessity or possibility) between the truth of some proposition or
the occurrence of some event and a certain set of circumstances or principles which
provide the contextual frame for the utterance. They do not, however, provide any
direct information about the actual identity of these circumstances or principles – they
are unmarked. Therefore, modal verbs can be ambiguous between different types of
modality and they can be vague with respect to, for example, the degree of speaker
involvement (subject-internal/subjective or subject-external/objective modality).
Periphrastic modal expressions, on the other hand, incorporate specific
information about one or more aspects of these variables. Consider the following
examples:
(79) Er muss morgen im Büro sein. „He must be in the office tomorrow.‟
(80) Mit Sicherheit ist er morgen im Büro. „He will certainly be in the office tomorrow.‟
(81) Ich bin überzeugt, dass er morgen im Büro ist. „I am convinced, that he will be in the office tomorrow.‟
(82) Er ist verpflichtet, morgen im Büro zu sein. „He is obliged to be in the office tomorrow.‟
(83) Ich verlange, dass er morgen im Büro ist. „I demand that he be in the office tomorrow.‟
(79) is the unmarked utterance with a modal verb. Without further context this is
ambiguous between epistemic or deontic reading, but also vague in terms of the
source of the necessity. If the speaker wishes to disambiguate the utterance, there
are various options. The modal expressions in (80) and (81) are explicitly epistemic,
the ones in (82) and (83) explicitly deontic. Within this distinction, (80) and (82)
60
additionally convey that the source of the (respective epistemic and deontic)
necessity is speaker-external (objective), while (81) and (83) identify the source of the
necessity as speaker-internal (subjective).
On top of this the speaker has the option of further modifying certain expressions
to achieve a mediating or intensifying effect, e.g.
(84) Ich bin fest davon überzeugt, dass er morgen im Büro ist. „I am absolutely convinced, that he will be in the office tomorrow.‟
Furthermore, because periphrastic modal expressions are less integrated into the
syntactic structure of a sentence as modal verbs, they offer the possibility of
positioning the part that contains the modal element in different places. A modal
adverbial, for instance, can be placed in sentence-initial, interpolating or sentence-
final, adjoining position:
(85) Meiner Meinung nach hat er es getan. ‟In my opinion, he did it.‟
(86) Er hat es meiner Meinung nach getan.. ‟He did, in my opinion, do it.‟
(87) Er hat es getan, meiner Meinung nach. ‟He did it, in my opinion.‟
On top of this, the fact that the modal element can be expressed in different
phrase types (e.g. noun phrases, adjective phrases, prepositional phrases etc.)
means that the same modal meaning can be incorporated into the sentence structure
in different grammatical ways. Both of these factors have bearings on the information
structure of an utterance, i.e. what is the topic and the focus of the utterance, which
pieces of the proposition are depicted as given and new information. There is not
enough space here to go into details about information structure theories42, but it
should be evident that a sentence like (88) below, has a different information
structure from the sentence in (89). In (88) the modal element is extrapolated from
the rest of the proposition in a separate, sentence-initial, clause (which actually
depicts the modal element as a focus of its own with the dass-clause containing a
second focus), while in (89) the modal element is incorporated into the clause as a
42
For a detailed account of information structure, see e.g. Lambrecht (1994). He defines information structure as a component of grammar that governs the relationship between the formal structure of sentences and the communicative situations in which sentences are used to convey pieces of propositional information. “In this information-structure component, propositions, as conceptual representations of states of affairs, undergo pragmatic structuring according to the discourse situations in which these states of affairs are to be communicated. The pragmatic structuring of propositions is done in terms of a speaker‟s assumptions concerning the hearer‟s state of mind at the time of an utterance. Pragmatically structured propositions are then paired with appropriate lexicogrammatical structures.” (Lambrecht 1994:334).
61
sentence adverbial in middle position (which makes the modal element only a part of
the single focus in the sentence).
(88) Es besteht die Möglichkeit, dass er den Diebstahl begangen hat. „There is a possibility, that he has commited the theft.‟
(89) Er hat den Diebstahl möglicherweise begangen. „He has possibly commited the theft.‟
All of these factors contribute to the polarity of modal verbs and periphrastic modal
expressions. The modal verbs are semantically generic and as part of the predicate
they are also syntactically more integrated than other modal expressions. These, in
turn, are often much more specific with respect to one or more of the variables just
discussed.
From a contrastive point of view, it has been stated that, in German, epistemic
modality is more often expressed by modal adverbials than by modal verbs, whereas
the opposite is the case for English, as the modal verb occupies a more central space
in epistemic modality here (cf. Edmondson et al. 1977:256; Leech and Svartvik
1975:292; Lyons 1977:802). Nehls (1986:176ff) mediates this statement slightly as
his own investigation43 reveals that there are different tendencies with respect to the
various levels of certainty of epistemic utterances. He found that for the expression of
confident conclusions as well as possibility, German indeed tends to prefer modal
adverbials such as sicher, gewiss (surely, certainly) and vielleicht, womöglich
(maybe, perhaps, possibly), respectively, whereas in English the modal verbs are
preferred over the modal adverbials. For the expression of confident assumptions,
however, there is no preference of modal verbs (WILL, SHALL) in English and modal
adverbials are used frequently in both languages (wahrscheinlich – probably).
Overall, he confirms that generally modal verbs are preferred over modal adverbials
for the expression of epistemic modality, but the difference between German and
English modal verb/modal adverb relationships is of a gradual rather than categorical
nature. On a general note, we can certainly state that, while English and German
have many modal expressions in common, the two languages differ with respect to
which of these modal expressions are preferred in certain contexts (cf. Heinrichs
1981).
43
Based on the contrastive analysis of the German novel “Die Blechtrommel” (Günter Grass) and its English translation.
62
2.3.5 Subjunctive mood (Konjunktiv)
Unlike English, the German language has two fully fledged, (in parts) morphologically
distinct, paradigms for subjunctive (Konjunktiv) moods, sometimes misleadingly
called Konjunktiv Präsens (present subjunctive) and Konjunktiv Imperfekt (past
subjunctive). These terms are used because the present subjunctive is formed on the
basis of the present tense stem of the verb, whereas the past subjunctive uses the
past tense stem of the verb. This is confusing, however, as the two subjunctives differ
with respect to modality, not tense (cf. e.g. Öhlschläger 1984:239; Brinkmann
1971:362). It is therefore preferable to refer to the two types as Konjunktiv I and
Konjuktiv II. In the following, I will first provide an outline of the basic functions
(“Grundleistung”, Brinkmann 1971:373) of the two subjunctive moods in German and
then explain in more depth how the respective morphological options for the
subjunctive are used in the German language today. The first part is based on
Brinkmann‟s (1971) account of the modal system, the second part draws on Zifonun
et al. (1997) and other relevant works on contemporary German grammar (e.g.
Helbig 2007).
2.3.5.1 The basic functions of Konjunktiv I and Konjunktiv II
Brinkmann (1971:366ff) explains the relationship between the different moods in the
following way: The indicative is the default mood for the assertion of information, i.e.
the indicative is used to signal that the utterance is about information. As such it
stands in opposition to the imperative – the default mood for implementations. It
signals that an utterance is aimed at prompting the conversational partner to carry out
the action expressed in the utterance. The subjunctives are, in principal, available for
both types of communication and therefore either contrast the indicative or the
imperative. They do, however, play a much bigger role in the realm of information.
In order to understand the basic capacity of the subjunctives, it is necessary to
consider again Brinkmann‟s notion of Horizont, which has already been mentioned in
the section on modal verbs (cf. Footnote 29, p. 39). The horizon incorporates all the
general and situational knowledge that people bring to a communicative situation,
including “their relationship to each other, their social roles, their experiences,
memories and expectations, of course also all their previous knowledge, everything
63
they think about”44 (Brinkmann 1971:730). A speaker‟s horizon is what they perceive
as valid and given in „their world‟. This world is the „real world‟ for this one person in a
particular situation, but is not necessarily congruent with other people‟s horizons or
even the same person‟s horizon in a different situation and therefore cannot be
accounted for or verified in terms of truth or reality. Brinkmann (1971:369) phrases
this in the following way:
“As with the tense system, there is no objective reference system. What is
taken as the horizon is established anew in every conversation on the grounds
of the particular conversational situation, the relationship between the
conversational partners and through everything that these partners bring to the
situation.”45
As mentioned before, the indicative (or imperative when an utterance is about
implementation) is the default mode for everything that falls within a speaker‟s
horizon. This can include a goal that is yet to be achieved or a condition that does not
yet exist in the actual world, as long as the speaker wishes to encode the proposition
as unmarked for the characteristic „outside my horizon‟.
According to Brinkmann (1971) the subjunctives, then, modify an utterance in two
different ways with respect to the speaker‟s horizon:
Konjunktiv I expands the given horizon by the introduction of a third party (in addition
to speaker and hearer), who is not present in the current conversational situation.
This affects the two types of communicative aims (implementation or information) in
the following way: For implementations, where the imperative is the default mode,
Konjunktiv I closes a systematic gap: The imperative can only be directed at persons
present in the conversation. As the Konjunktiv I introduces a third party into this
situation, a prompt can be directed at this third person. This usage of the Konjunktiv I,
however, sounds archaic and is nowadays only found in literary language and
formulaic phrases. In everyday speech these forms are now replaced by the modal
verbs SOLLEN or MÖGEN.
44
“Es schließt ihr Verhältnis zueinander ein, ihre soziale Rolle, ihre Erfahrungen, Erinnerungen und Erwartungen, natürlich auch ihr „Vorwissen“, alles woran sie denken.“ (Brinkmann 1971:730))
45 "Wie beim Tempussystem besteht also [beim Modus und dem Modalverbsystem] kein objektives Bezugsystem. Was als Horizont gegeben ist, wird jedesmal neu bestimmt durch die jeweilige Situation, durch das Verhältnis der Partner zueinander und durch das, was die Partner in die Situation mitbringen."
64
(90) Noch ist es Tag, da rühre sich der Mann... (Goethe) „It is still day, so man should still be working…‟
Konjunktiv I plays a much more important role in the realm of information. Here, it
is grammaticalised as the mode for indirect speech, as it enables the speaker to
explicitly mark an utterance as that of somebody else who is not present in the
current communicative situation. But using Konjunktiv I in indirect speech makes no
claim about whether the speaker also commits to this proposition or not.
(91) Er sagte, er sei zu krank um zur Arbeit zu gehen. „He said he was too ill to go to work.‟
Konjunktiv II goes one step further than expanding the horizon insofar as it marks
a proposition as situated OUTSIDE the given horizon. It is the mode for non-
factuality. When used in the realm of implementations, Konjunktiv II expresses
wishes that, from the perspective of the speaker, are unattainable (beyond the given
horizon):
(92) Wäre ich doch noch einmal jung! „If only I was young again!‟
This use is closely linked to conditional clauses in informational contexts (as the use
of „if‟ in English in these cases also indicates). The use in conditional clauses is the
most prominent „operational sphere‟ of the Konjunktiv II. Brinkmann (1971:377)
explains that in conditional clauses the Konjunktiv II stands in distinctive opposition to
the indicative and that it is only the speaker‟s horizon, which is the deciding factor.
Consider the following two sentences:
(93) Wenn ich Zeit habe, komme ich dich besuchen. „If I have time, I‟ll come to visit you.‟
(94) Wenn ich Zeit hätte, käme ich dich besuchen. „If I had time, I would come to visit you.‟
In both sentences the condition (protasis) is a state that does not exist in reality – it is
hypothetical in both cases. Nevertheless, this hypothetical supposition can be put in
the indicative or the Konjunktiv II. The difference is that in (93) the supposition is
declared as within the speaker‟s horizon (i.e. it is in accordance with the speaker‟s
knowledge that the condition can be fulfilled). In (94) on the other hand, the
Konjunktiv II explicitly indicates that the condition is in conflict with the speaker‟s
horizon, in other words it expresses a condition that is only attainable if the speaker
could break out from the given horizon (which is, by definition, impossible). Indeed,
the fact that the condition is unattainable is probably the motivation for the utterance.
65
In a general note, Brinkmann (19971:379) recognises “the importance of this
linguistic breakout of the given horizon [through the subjunctive, but also e.g. modal
verbs], as it attests the freedom of human beings to mentally step out of the
boundaries of the immediate situation”46. He insists that “the crucial point is to regard
reality not as irrevocably determined, but to explore new possibilities. The
grammatical form for such explorations is the Konjunktiv II”47 (Brinkmann 1971:380).
It is this quality of the subjunctive that makes it relevant for argumentative writing.
So far, Brinkmann‟s account of the subjunctive within the modal system gives the
impression of neat and straightforward form-function relationships for both Konjunktiv
I and Konjunktiv II. The linguistic reality, however, is far more complex. Brinkmann
(1971:362) summarises: “The subjunctives […] are [to a large extent] morphologically
indistinct, they are only grammatically compulsory in a very limited range of contexts
and, because of their reference to the (very diverse) horizon, they are also
semantically only vaguely determined.”48 The subjective nature of the horizon has
already been discussed. This leaves two fundamental issues to be clarified: The
extensive syncretism between subjunctives and indicative (plus the so-called
“replacement system” including the analytic subjunctive with würde) and the contexts
in which subjunctives are mandatory or optional. As Zifonun et al. (1997:1731ff)
provide a much more detailed and up-to-date description of both of these aspects,
the following sections are based on their explanations.
2.3.5.2 Syncretism between subjunctives and indicative
Due to space restrictions, the problem of morphological syncretism between
subjunctive and indicative forms is only laid out along very general lines; for a
comprehensive account see Zifonun et al. (1997:1739). Konjunktiv I and Konjunktiv II
are formed by adding the subjunctive morphemes (-e, -est, -e, -en, -et, -en) to the
present tense verb stem and the preterite verb stem, respectively. This has the
following consequences:
46
“Das Ausbrechen aus dem gegebenen Horizont spielt in der Sprache eine große Rolle, weil sich darin die Freiheit des menschen bekundet, die Grenzen der jeweiligen Situation im Geiste zu überschreiten.“
47 “Es geht darum, die Wirklichkeit nicht als endgültig gegeben anzusehen, sondern neue Möglichkeiten zu erproben. Die grammatische Form solcher Erprobung ist der Konjunktiv II.”
48 “Allerdings sind die Konjunktive morphologisch […] undeutlich, nur sehr eingeschränkt in bestimmten Verwendungsweisen grammatisch verbindlich und durch den Bezug auf den (sehr verschiedenen) Horizont nur vage festgelegt.”
66
Depending on the phonological structure of the verb and whether a verb is strong
(irregular) or weak (regular), Konjunktiv I and indicative can sometimes be
distinct only in the 3rd person singular:
Indicative Konjunktiv I
1st singular ich arbeite ich arbeite
2nd singular du arbeitest du arbeitest
3rd singular sie arbeitet sie arbeite
1st plural wir arbeiten wir arbeiten
2nd plural ihr arbeitet ihr arbeitet
3rd plural sie arbeiten sie arbeiten
TABLE 3: PRESENT TENSE INDICATIVE AND KONJUNKTIV I
For other verbs, more forms of the paradigm are distinct, but the 1st person
singular and plural and the 3rd person plural are always homonymous:
Indicative Konjunktiv I
1st singular ich gebe ich gebe
2nd singular du gibst du gebest
3rd singular sie geben sie gebe
1st plural wir geben wir geben
2nd plural ihr gebt ihr gebet
3rd plural sie geben sie geben
TABLE 4: PRESENT TENSE INDICATIVE AND KONJUNKTIV I
For Konjunktiv II syncretism with the indicative preterite exists for all weak verbs
in all persons:
Indicative Konjunktiv II
1st singular ich spielte ich spielte
2nd singular du spieltest du spieltest
3rd singular sie spielte sie spielte
1st plural wir spielten wir spielten
2nd plural ihr spielte ihr spielte
3rd plural sie spielten sie spielten
TABLE 5: PRETERITE INDICATIVE AND KONJUNKTIV II (WEAK VERBS)
Strong verbs, which undergo stem vowel change in the preterite, have Umlauts
wherever possible in the Konjunktiv II form and are therefore always distinct from
the indicative:
67
Indicative Konjunktiv I
1st singular ich kam ich käme
2nd singular du kamst du käm(e)st
3rd singular sie kam sie käme
1st plural wir kamen wir kämen
2nd plural ihr kamt ihr käm(e)t
3rd plural sie kamen sie kämen
TABLE 6: PRETERITE INDICATIVE AND KONJUNKTIV II (STRONG VERBS)
All other periphrastic forms of the paradigm (future, perfect etc.) are formed, as in
the indicative, through combinations of the finite subjunctive verb forms of
sein/haben and werden and the participle/infinitive of the main verb. Out of the
copula/auxiliary verbs (sein, haben, werden), sein is the most distinct in
Konjunktiv I, all of them are distinct in Konjunktiv II:
Indicative Konjunktiv I Konjunktiv II
ich bin/war ich sei ich wäre
ich habe/hatte ich habe ich hätte
ich werde/wurde ich werde ich würde
TABLE 7: SEIN/HABEN/WERDEN IN INDICATIVE, KONJUNKTIV I AND KONJUNKTI II
Because of this varied and extensive syncretism, it is often not possible to
determine whether a verb in a particular utterance is in the subjunctive or the
indicative. There is therefore a complex replacement system in place for cases where
the speaker explicitly wants to mark a verb as subjunctive. This used to be portrayed
in the following way: If syncretism between indicative and Konjunktiv I occurs (95),
Konjunktiv II is used instead. If the Konjunktiv II form is homonymous with the
indicative preterite (96), another so-called „replacement subjunctive‟ is used (97). This
is formed analytically with the Konjunktiv II of werden – würde – and the bare
infinitive:
(95) Die Dozenten sagen, dass die Studenten zu wenig arbeiten. =
(96) Die Dozenten sagen, dass die Studenten zu wenig arbeiteten. =
(97) Die Dozenten sagen, dass die Studenten zu wenig arbeiten würden. „The lecturers say that the students work too little.‟
The real situation, however, is more complicated. Bausch (1979:114) argues
convincingly that it is “concepts of language cultivation, which dominate linguistic
criticism, [that] have made it into modern grammars as the desirable standard in the
68
description of modality”49, especially where an insistence on synthetic subjunctive
forms and the rejection of the würde-subjunctive is concerned. Zifonun et al.
(1997:1783ff) summarise on the basis of several corpus investigations (including
Bausch‟s (1979)) the actual situation of synthetic and analytic subjunctives in today‟s
language use:
Konjunktiv I is generally restricted to literary texts and public (in particular mass-
media) communication. While in literature, the use of Konjunktiv I is connected to
genre- and author-specific conventions and certain literary effects, in mass-media
communication and in particular in news texts, indirect speech is always marked by
the subjunctive. In everyday spoken discourse, Konjunktiv I hardly occurs at all and if
so, usually only on the verbs haben and sein. The other (rare) instances of the
Konjunktiv I in spoken language can usually be interpreted as interferences from
written language. Instead, Konjunktiv II, würde-subjunctive and indicative50 are used
in indirect speech contexts.
Konjunktiv II is used in spoken and written language and is therefore less
dependent on text type, genre, language variety or speech community. The
differences here are with respect to which verbs are used in the Konjunktiv II. In
spoken language, only haben, sein, werden and the modal verbs are used almost
exclusively with the synthetic Konjunktiv II. Most other main verbs (with the exception
of some very common ones such as kommen, gehen, wissen, geben, tun, brauchen)
are usually used with the analytic würde-subjunctive, even if a distinct Konjunktiv II
form is available. In written language, the range of verbs that occur with synthetic
Konjunktiv II is slightly wider, but there is also a tendency, just as in spoken
language, to restrict its use to certain common verbs and use the würde-subjunctive
in all other cases (cf. Witton 2000). In light of these findings, Zifonun et al.
(1997:1785) suggest that what in terms of the morphological paradigm is taken as the
„replacement subjunctive‟, i.e. the würde-subjunctive, in terms of its frequency of use
in spoken AND written language actually occupies the central position. Bausch
(1979:61) goes as far as to assess the different subjunctive forms as stylistic
variations, where Konjunktiv I is assigned only to written language and the würde-
form is structurally the main variant and most frequently used variant of the Konunktiv
II. Fabricius-Hansen (1997) predicts that “the würde-subjunctive will probably, over
49
“… die dominierenden sprachpflegerischen Konzepte der Sprachkritik in neuere Grammatiken weitgehend als Prämissen in die Beschreibung des Modusbereichts eingegangen sind.”
50 N.B. a subjunctive marker is not grammatically compulsory in indirect speech contexts. This will be discussed in more detail in the next section (2.3.5.3).
69
time, entirely replace the analytic Konjunktiv II forms and will in the end possibly be,
with a few auxiliary exceptions, the only Konjunktiv II forms.”51
2.3.5.3 Usage contexts for Konjunktiv I and Konjunktiv II
The final point that needs addressing in the discussion of the subjunctive in German
is that while it is only grammatically compulsory in a small number of contexts it can,
of course, occur in a much wider variety of context types. I will, again, follow Zifonun
et al.‟s (1997:1743) comprehensive description of these contexts types here (cf. also
Helbig 2007). Zifonun et al.‟s definition of the basic function of the subjunctive
corresponds with Brinkmann‟s (1971) description outlined above: “The subjunctive
indicates a break or annulment of the immediacy in the interpretative reference with
respect to two primary coordinates: present speaker and factual world”52 (Zifonun et
al. 1997:1785)(emphasis in original). In light of the results on actual language use
summarised above, however, they state that, while each subjunctive mood can be
primarily ascribed to one of the two functions (i.e. Konjunktiv I for indirect speech,
Konjunktiv II for non-factuality), both subjunctives can reach into each other‟s domain
(especially Konjunktiv II in indirect speech)53. Zifonun et al. (1997:1743) term the two
groups of context types that arise from the two basic functions of the subjunctive
„indirectness contexts‟ and „modality contexts‟. As these context types are crucial for
the analysis of the subjunctive in the present study, they will be summarised here.
First of all, they ascertain that the subjunctive is the marked mood for indirectness
and non-factuality contexts; it excludes directness and factuality contexts. The
indicative, on the other hand, is unmarked for these contexts, i.e. it can generally
occur in all contexts. In these cases, the default interpretation is directness and
factuality unless the context signals the transfer into an indirectness or non-factuality
context.
a) Modality contexts
Modality contexts explicitly or implicitly indicate that a proposition is not to be related
to what is factual. Certain modality contexts even infer that the proposition is
51
“Es steht aber zu vermuten, daß die würde-Formen mit der Zeit generell den Platz als analytische Konjunktiv-II-Formen einnehmen und am Ende möglicherweise, bis auf vereinzelte auxiliare Ausnahmen, die einzigen Konjunktiv-II-Formen überhaupt sein werden.”
52 “Der Konjunktiv zeigt eine Brechung oder Aufhebung der Unmittelbarkeit der interpretativen Bezugnahme auf die beiden primären Koordinaten aktualer Sprecher oder aktuale Welt an.”
53 Cf. also Fabricius-Hansen (1997:20) for a similar summary of the main and peripheral functions of Konjunktiv I and II and areas where their use overlaps without changing the meaning or grammaticality of an utterance.
70
decidedly non-factual (counterfactual). Where a subjunctive (Konunktiv II or würde-
form) is used this signals the following semantic restriction [A] (Zifonun et al.
1997:1744):
[A] ‟It is in accordance with the speaker‟s knowledge, that the piece of prepositional
knowledge which is formulated using the subjunctive is not the case.‟ In
counterfactual situations this restriction is tightened to: „It is not in accordance with
the speaker‟s knowledge that the proposition is the case.‟ The interpretation as non-
factual or counterfactual is dependent on context.
Conditional clauses
The most common kind of modality context is conditionals. Syntactically, the
condition is realised by a subordinate clause introduced by wenn, falls, sofern (if) or a
verb-first subordinate clause; the würde-form can occur in subordinate and/or main
clause with or without future reference.
Semantically, the semantic restriction [A] formulated above applies, first of all, to
the condition (p) – which sets up the interpretation as non-factual or counter-factual.
On top of that the following relationship [B] between condition (p) and consequence
(q) applies:
[B] ‟According to the speaker‟s knowledge it is more probable that (p and q) is the
case than (p and not q)‟. This is regardless of whether p is perceived as never true or
possibly attainable (Zifonun et al. 1997:1746).
(98) Wenn er krank wäre, könnte er die Prüfung nicht ablegen. ‟If he was ill, he couldn‟t take part in the exam.‟
Counterfactual conditionals express that, from the speaker‟s point of view, it is not the
case that p and q are both the case at the same time. They often refer to events in
the past (retrospective counterfactual conditionals):
(99) Wenn er mich rechtzeitig gewarnt hätte, hätte ich diesen Fehler nicht gemacht. ‟If he had warned me in time, I wouldn‟t have made this mistake.‟
The relationship [B] also applies to conditionals in the indicative. The semantic
restriction [A], however, is then to be replaced with the following restriction [C]:
[C] „It is in accordance with the speaker‟s knowledge that the proposition formulated
in the condition (p) is the case.‟ (Zifonun et al. 1997:1746)
(100) Wenn sie heute ankommt, hole ich sie vom Bahnhof ab. ‟If she arrives today, I will pick her up from the station.‟
This means, as already pointed out above, that conditionals formulated in the
indicative are still hypothetical arguments (which is why the often used term „real
71
conditions‟ is inappropriate) as both p and q refer to the future, but here the
probability that p is the case is emphasised. I will therefore work with Nehls‟ (1978)
terminology here and call this type „direct prospective predicative conditionals‟.
Conditionals of the type exemplified in (98) above (Konjunktiv II in both clauses in
German) are called „indirect prospective predicative conditionals‟ as p and q still both
refer to future events but the propositions are expressed more tentatively (indirect).
Mixed forms are also possible, where one of the two clauses contains the indicative,
the other one the subjunctive:
(101) Wenn auf diese Option verzichtet wird, wären zwei weitere Möglichkeiten zu bedenken. ‟If this option is abandoned, two other possibilities would have to be considered.‟
Potential/non-factual argumentation
These are contexts where other possible (but not factual) circumstances are set up,
under which a non-factual proposition can be viewed. Syntactically this is achieved
e.g. by noun phrases, participle constructions, infinitive constructions or adverbials
which create scenarios by either referring to something as existing just as a thought
or referring to things that are not factual or depicting situations that are not given in
the here and now.
(102) Er hat großes Glück gehabt. Ein schlechterer Schwimmer wäre ertrunken. „He was very lucky. Someone who is a worse swimmer would have drowned.‟
A further series of modality contexts are triggered by certain lexical or semantic
markers in a super-ordinate clause:
Counterfactual consecutive clauses
These have indicative in the main clause and the particle so/zu/allzu (too), while the
subordinate clause is introduced by (als) dass:
(103) Das Wasser ist zu kalt, als dass man darin baden könnte. The water is too cold, as that one in it swim could. „The water is too cold to swim in it.‟
Other variations of this construction are clauses introduced by ohne dass and
(an)statt dass.
Counterfactual comparative clauses
These also have indicative in the main clause, while the subjunctive clause is
72
introduced by als/als ob (as if). With these constructions Konjunktiv I or Konjunktiv II
can be used.
(104) Die Unterschrift kam mir keineswegs so vor, als ob sie gefälscht sei. „The signature didn‟t seem to me at all as if it was fake.‟
Counterfactual relative clauses
In relative clauses following a negated main clause, Konjunktiv II can be used to
indicate that the relative clause is within the scope of negation.
(105) Es war niemand auf der Party, der sie verstanden hätte. „There was nobody at the party who would have understood her.‟
Modality contexts with modal verbs
When used in conjunction with modal verbs, Konjunktiv II is often used to signal the
non-factuality of the whole proposition including the modal verb.
(106) Sie hat immer über ihre Verhältnisse gelebt. Das hätte sie nicht machen sollen. „She has always lived beyond her means. She shoudn‟t have done that.‟
This works in the same way with verbs like wünschen, ermöglichen, verdienen, sich
lohnen (wish, enable, deserve, be worth) with infinitive or dass-clause.
(107) Er wünschte sich so sehr, dass sie ihm nur einmal richtig zuhören würde. „He wished so much that she would, just once, really listen to him.‟
Modality contexts for politeness
These contexts include e.g. attitude statements or discourse structuring statements
but also often occur in sales situations. The point here is that the utterance is only
seemingly non-factual in order to reduce the interactional immediacy.
(108) Nach diesem Exkurs wäre es jetzt an der Zeit, dass wir uns wieder auf das eigentliche Thema konzentrieren. „After this excursion it would now be time for us to concentrate on the actual topic.‟
b) Indirectness contexts
In indirectness contexts the speaker indicates that the propositional content they are
expressing does not relate to their own knowledge at the time of speaking, but refers
to a different source. This is commonly referred to as indirect speech of an actual or
imagined original utterance. This has obvious consequences for transferring
personal, spacial and temporal deictic expressions as well as capturing in an
appropriate reporting verb the type of speech act that the original utterance intended.
These issues are not of interest here. The relevant aspect for the use of the
73
subjunctive is the directness/indirectness distinction. This refers to the degree of
immediacy that the speaker commits to when making a statement. Directness is
associated with the claim „I say that p‟. Indirectness is associated with a moderated
commitment, as in: „X says that p, but I leave it open whether I say that p‟, or with a
complex commitment (a combination of the previous two), as in „X says that p and I
also say that p.‟
The difficulty here is that in German there is no 1:1 assignment of indicative and
subjunctive to these different degrees of commitment, as the indicative can occur in
any one of these cases. As explained before (2.3.5.2), the use of the subjunctive in
indirectness contexts is highly genre, discourse and text type dependent. We can
distinguish between discourse/text types etc. where indirectness is usually explicitly
marked following the normative directives of prescriptive grammar and discourse/text
types where indirectness is not usually explicitly marked. The former typically occurs
in public communication (esp. mass-media news texts). In these, indirectness is at
least signalled by one marker, either (i) a subordinate clause with a reporting verb in
the main clause or (ii) the subjunctive, but often double marking occurs. Here, the
subjunctive is the „default mode‟ for indirect speech. In non-public, informal
communication, on the other hand, indirectness is not usually marked. Here, the
indicative is the default mode, even in indirectness contexts that express the
moderated commitment of the kind „X says that p, but I leave it open whether I say
that p‟.
So, in summary, the use of the subjunctive always explicitly signals moderated
commitment. If the indicative is used in indirect speech it is entirely text or discourse
dependent whether this to be interpreted as directness (claim) or indirectness
(moderated commitment) or a combination of both (complex commitment).
It should have become clear that the use of the various subjunctive forms in
German is mostly subject to factors other than grammatical rules. These factors
include genre, discourse and text type, but also language variety and individual
speaker choices (cf. also Lauridsen and Poulsen 1995). In the present study on
learner German, the contexts in which the learners employ subjunctives should
therefore give an indication on how their language use fits into the specific
conventions on the use of subjunctives in student argumentative texts.
This concludes the description of the diverse range linguistic possibilities for
expressing modality. Each of the four groups of modality markers – modal verbs,
periphrastic modal expressions, modal infinitives and subjunctive moods – add
slightly different layers of modality to an utterance, as they all differ in their
74
morphological and syntactic make-up and integration into the sentence. As disparate
as they are, however, all modal means express the same fundamental function: Any
form of explicit modality marking by any of the means discussed above indicates that
the validity of an utterance is in some way or form mediated, i.e. the utterance is to be
interpreted with respect to the kind of modality that is encapsulated in the modality
marker and not to be questioned in terms of the truth value of the expressed
proposition.
75
3. Modality in Second Language Acquisition Research
As pointed out in section 2.1.3, studies of modality in general have mainly
concentrated on modal verbs. The same holds true for investigations of modality in
first and second language acquisition. One reason might be the relative ease of
investigating modal verbs, as they form a closed set of items. Another might be that,
without the possibilities of corpus methods, an investigation beyond modal verbs
would become unfeasible. While this restriction on the research focus is therefore
understandable, it nevertheless also limits the informative value of those studies. This
is even more the case if we consider the fact that, although different languages may
have the same types of modal expressions at their disposal, they often differ
considerably with regard to which of these expressions are preferred in comparable
contexts (Heinrich 1981:74).
There are, however, some qualitative studies that look very closely into the
emergence of modal markers, including not only lexical and grammaticalised means
of expressing modality but also implicit uses of modality via inference from the
discourse context (Giacalone Ramat 1999), as well as 'precursors' of
grammaticalised modalities (Dittmar 1993) in natural language acquisition settings.
3.1 MODALITY IN NATURALISTIC ACQUISITION
Giacalone Ramat (1995; 1999) investigates the order of emergence of (epistemic and
deontic) modal distinctions in learner varieties and the types of encoding of modal
notions preferred by adult learners of Italian. The subjects in her study (from the
longitudinal “Pavia Project”) comprised 25 learners with different L1s, who learnt
Italian in a natural acquisition setting in Italy without formal instruction. The key point
in her results is that “the acquisition of linguistic means for expressing deonticity and
epistemicity show quite different pictures; while deonticity is straightforwardly
expressed by the modal verbs […], epistemic meanings are first conveyed by lexical
strategies, such as propositional attitude verbs, adverbs, adjectives." (Giacalone
Ramat 1999:394).
Evidence from the P-MoLL project (Modalität in Lernervarietäten im Längsschnitt -
Modality in a longitudinal study of learner varieties) suggests that the same can be
observed for learner German. P-MoLL is a longitudinal research project on the
acquisition and grammaticalisation of modality conducted by Dittmar et al. (see
76
Dittmar, Reich, Schumacher, Skiba and Terborg 1990) with Italian and Polish
learners starting at elementary proficiency levels in untutored language learning
settings. Within this project, Dittmar and Ahrenholz (1995:206ff) are able to
demonstrate that propositional attitude verbs such as denken, glauben (think,
believe) and adverbs like vielleicht (perhaps) dominate epistemic modality in the early
stages of acquisition while epistemic readings of modal verbs occur later and less
frequently. This seems to be a language-independent feature of learner language, as
similar results could also be attested for French learners of English (Lambert 1995).
Terborg (1993) also shows that the first occurrences of modal verbs in learner
language are exclusively deontic. These deontic modal verbs occur early on in the
acquisition process and play a central role as deontic modal expressions for a long
time, while deontic adjectives and adverbials only occur much later and are rare
(Dittmar and Ahrenholz 1995). Giacalone Ramat (1999:381ff) is able to demonstrate
that this grammaticalisation process in the learner language from deontic to epistemic
meanings of modal verbs maps the historical grammaticalisation process of modal
verbs. On top of this, there is ample evidence that, in first language acquisition,
deontic meanings of modal verbs also precede epistemic ones (e.g. Perkins 1983;
Sheperd 1993; Stephany 1995; for the first language acquisition of epistemic
KÖNNEN versus vielleicht see Doitchinov 2001). Terborg (1993:245) partly assumes
the same underlying reason for this similarity: “[…] marked forms of deontic necessity
occur before marked forms of high degrees of probability in L1- and L2-acquisition,
because a speaker does not need to mark factivity but must mark deontic necessity,
if he wants to express it.”
In first language acquisition, another reason for the later emergence of epistemic
modality is ascribed to general cognitive development: “Young children‟s social status
as well as their physical condition favor their concern with norms for actions and the
possibility of performing them, the attainment of desired states of affairs with the help
of others, rather than the validity of statements” (Stephany 1995:106). This is where
adult second language learners differ from L1 child learners: Adult SLA learners
already have pragmatic knowledge of both deontic and epistemic categories but,
more than that – they also express both types of modality from early on and gradually
develop the linguistic resources in the target language to do so concurrently. Dittmar
(1993), Dittmar and Ahrenholz (1995) and Giacalone Ramat (1999) trace the
emergence of modality before target-like grammatical or lexical means are available
to the learners in the early learner stages. All of these studies show that although no
explicit modal expressions are present, modal meanings can still be conveyed
77
through implicit means, i.e. where “some kind of modalization has to be posited on
the basis of discourse context […] Modality is to be inferred from the (native)
interlocutor‟s utterances, or from discourse pragmatic features” (Giacalone Ramat
1999:387). The following utterance (adapted from Ahrenholz 2000:349) exemplifies
how in an instruction the learner (12 months of residence in Germany) explicitly
specifies object, goal and location referents but conveys the modal conditions – non-
factuality, necessity – only implicitly:
(109) mit de journal de aschenbäsch in de boors with the newspaper the ashtray in the bag „(Please put) the ashtray (with the help of) the newspaper into the bag‟
In addition to this so called “pragmatic mode” (Givón 1979), unanalysed formulaic
expressions (stereotypes), which are used in an invariant way and have some kind of
“proto-modal meaning” (Dittmar 1993:224) are employed in these early stages.
Ahrenholz (2000) shows how learners perform instructions in pragmatic mode by
using verbless directives or directives including formulaic, uninflected verb forms.
Dittmar (1993) demonstrates how the German word bitte (please) is used by the
learner in instructions to convey politeness and deontic necessity at the same time
(example from Dittmar 1993:226):
(110) bitte zeitung tasche „please newspaper bag‟ Please put the newspaper into the bag
Terborg (1993:242) adds that his observations indicate that bitte serves more as
an expression of necessity than of politeness. This is also supported by evidence
from the syntactic development in learner varieties from sentences containing bitte to
those containing the conventional modal verb müssen to express deontic necessity
(Skiba 1993).
The repertoire for expressing epistemic modality in basic learner varieties
comprises paralinguistic cues, such as hesitations or gestures, and intonation (e.g.
Dittmar 1993). As the acquisition progresses, formulaic expressions such as I don‟t
know (no so in Italian; weiß nicht in German) are added as markers of epistemic
possibility. Vielleicht (perhaps) is regarded as a kind of epistemic „joker‟, which is
used extensively either on its own or as reinforcement of epistemically used verbs of
sensory perception, thinking and believing, such as denken, finden, glauben (think,
find, believe), which mark a further step in the grammaticalisation process that
78
learners go through. Giacalone Ramat (1992) refers to this process as “acquisitional
grammmaticalization”, which is defined as
“The transformation of pragmatic and semantic formats of expression (words,
unanalyzed “chunks”)
1. Into specific productive syntactic patterns of the individual learner variety,
which are important for the organization of meaning and which tend to
converge in the long run with the target variety; and
2. Toward morphosyntactic norms of the target language.” (Skiba and Dittmar
1992:324)
This can be seen in the development from implicit modality in example (109) to
explicit marking with bitte in example (110) to the next step where deontic necessity
is expressed by the appropriate modal verb MÜSSEN as in example (111) below,
even if the whole utterance is not yet entirely target-like (example from Terborg
1993:240):
(111) du muss die Zeitung bisschen höher you must the newspaper a little higher „You have to (hold) the newspaper a little higher.‟
Where the acquisitional grammaticalisation of modal verbs is concerned, it could,
again, be shown that they emerge according to an order that is regularly followed by
learners, although this order differs between different L2 languages and L1-transfer
also seems to play a role here. Dittmar and Terborg (1991) determined that Polish
learners of German display deontic MÜSSEN (must) first, whereas for an Italian
learner WOLLEN (want) is the first modal verb to appear, followed by MÜSSEN and
KÖNNEN(can). Dittmar (1979) determines in his functional analysis of emergent
modal verbs that the high frequency of WOLLEN and MÜSSEN is also due to an
expansion of meanings where e.g. MÜSSEN is not only used for demands,
suggestions and advice but also for temporal meanings, ability and permission. While
these three modals (in their deontic and volitive meaning, respectively) are acquired
at an early stage, a second group of modals including SOLLEN (is supposed to),
MÖCHTEN (would like to) and DÜRFEN (be allowed to) emerges considerably later.
Dittmar and Ahrenholz (1995:231) add, however, that conceptual proximity to the
native language plays an important role. For the Italian learner WOLLEN is the
prototypical volitional expression, which corresponds both in syntactic form and
semantic function to the Italian VOLERE – it is acquired first. Polish learners, on the
other hand, develop MÖCHTEN before WOLLEN, because there is an analogy to
the corresponding Polish modal verb.
79
It has to be added here, though, that the data basis for these qualitative studies is
not very extensive and this is a major drawback. The Pavia Project comprises only 25
learners with various L1-backgrounds. This means that the data that is available for
speakers of the same mother-tongue is relatively small, although L1 is a significant
variable for differences in learner language. The P-MoLL project only comprises 16
learners, but for the studies reported on here that draw on this data, not even all of
the data is used. Dittmar and Terborg (1991) and Dittmar (1993), for example, only
analyse the recordings of one Polish learner. Ahrenholz‟s (2000) study is based on
the recordings of 10 learners. Despite the fact that the findings of the various studies
in the different projects seem to corroborate each other, we should be cautious of
overgeneralising these results.
3.2 MODALITY IN INSTRUCTED ACQUISITION
In all the evidence for the grammaticalisation process of adult learners presented so
far we have seen that, at the beginning of the acquisition process, pragmatic needs
are the driving force behind the development of lexical and grammatical means. The
learners “build on their pragmatic knowledge, making do with whatever L2 grammar
they have and at the same time acquiring the grammar needed to accomplish actions
in the L2” (Kasper and Rose 2001:188). This “primacy of pragmatics” hypothesis
(Dittmar 1992:254) therefore states that “pragmatic categories precede syntactic
ones” (cf. also Kasper and Rose 2002 for further evidence of this hypothesis in areas
other than modality). Conversely, there is rich evidence, especially from more
advanced L2 learners in instructional settings, where correct grammar is used in a
pragmatically inappropriate way. This is the so called “grammar-precedes-pragmatics
scenario” (Kasper and Rose 2001:174). This scenario comes in three different forms
(cf. Kasper and Rose 2001:190):
Firstly, learners demonstrate knowledge of a particular grammatical structure or
element but do not use it to modify illocutionary force. Secondly, they demonstrate
knowledge of a grammatical structure but use it to express pragmalinguistic functions
that are not conventional in the target language. And thirdly, learners demonstrate
knowledge of a grammatical structure and its pragmalinguistic functions yet use it in
inappropriate sociopragmatic contexts.
The first form essentially says that grammatical knowledge does not necessarily
enable appropriate pragmalinguistic use (utterances which convey the intended
80
illocutionary force) and is well documented in two studies of learner modality. In a
one-year longitudinal study with twelve participants in an intensive instructional
programme, Salisbury and Bardovi-Harlig (2000) examined the emergence of
epistemic modality in beginning English as a second language (ESL) learners with
varying L1s. Despite considerable variability in the learners‟ modality profiles, modal
expressions emerged in a consistent acquisitional pattern:
maybe > think > CAN > WILL > WOULD > COULD
As with the learners in natural settings, we can see that adverbials and
parenthetical expressions precede modal verbs in epistemic modality, but the crucial
outcome of this study is that even the availability of modal verbs in the learners‟
grammars did not necessarily translate into the use of these verbs in pragmatically
appropriate ways (in this case for mitigating disagreement). Salsbury and Bardovi-
Harlig (2000:73) observe that “even learners with grammaticalized expressions of
modality rely heavily on lexical forms to unambiguously mark their pragmatic intent”.
This corroborates evidence from an earlier cross-sectional corpus study on epistemic
modality by Finnish learners of English conducted by Kärkkäinen (1992). She found
that her lower-proficiency learners preferred to modify their speech acts by
parentheticals and lexical verbs (mainly I think) as a compensatory strategy for both
modal verbs and adverbs. The higher-proficiency learners chose to use modal verbs
and adverbs more frequently, though not as much as the native speakers. Their
preferred epistemic options were still parentheticals. Both of these sets of evidence
seems to suggest, then, that syntactically integrated, non-routinised expressions of
epistemic modality (such as the modal verbs and adverbs) – which convey speaker
attitudes implicitly – are more difficult to acquire than explicit, extra-clausal and
routinised expressions, such as the parentheticals. Kärkkäinen (1992:213) suggests
that this implicitness makes it difficult to pinpoint and describe the exact pragmatic
functions of these expressions, which in turn means that they are not explicitly taught
and can evade the learner‟s attention for a very long time.
To sum up the relationship between grammatical and pragmatic L2 acquisition, we
have seen that, while the direction from pragmatic categories to grammatical
expressions characterises the very early stages of (mainly untutored) L2 acquisition,
the reverse seems to be the case for more advanced learners (and, we might add,
learners in instructional settings). Their learning task increasingly changes to figuring
out the various pragmatic, often secondary meanings that specific grammatical forms
(e.g. modal verbs) have.
81
So far, all the studies I have discussed investigated the spoken language of
beginning to mid-proficiency learners, and considered mainly the involvement of
pragmatics on grammatical development. The selection of beginning learners –
especially in natural acquisition settings – explains the focus on spoken language in
these studies, as this group of learners would not produce much written text in the L2
in their everyday lives and asking them to do so for a research project would bring in
additional issues and complications that would negatively influence the validity of the
results. The study presented within this thesis, however, deals with mid- to high-
proficiency university students of German who produce written texts in the L2
regularly as part of their degree course. As discussed before (see 1.1), the
argumentative texts that the learners are asked to write as part of their assessment is
one type of text where expressions of modality are essential. Unlike the beginning
learners in the studies discussed previously, the university students are expected to
have already acquired a substantial part of the modal system with its various
linguistic manifestations and have these modal expressions available for both spoken
and written discourse. How they employ these expressions in comparison to native
speakers is therefore the focus of this study, rather than the development of early
forms of modality. In the following part of this review I shall therefore include two
modality studies on advanced learners in instructional settings and written discourse.
These are particularly relevant here, as both medium and learner proficiency matches
more closely the study presented in this thesis. Their design, results and discussion
of findings will be presented separately in slightly more detail. On top of this, other
corpus studies on learner modality are discussed very briefly.
The first investigation is a corpus study by Aijmer (2002) involving Swedish
learners of English at Swedish universities, who have already gained a relatively
advanced level of proficiency. The data forms part of the ICLE (International Corpus
of Learner English) corpus collected at the university of Louvain (for a full description,
see Granger 1998) and consists of a 52,000 word sample of the 200,000 word
Swedish subcorpus of argumentative written texts. Recognising that, despite the
central role of modal verbs in the English language, a wide variety of alternative
modal expressions exists and needs to be taken into account, especially when
investigating learner language, Aijmer seeks to investigate the range and frequency
of some key modal verbs and (epistemic) adverbs in the learner corpus. She does
not, however, consider adjectival or noun phrases in her analysis. The learner data is
compared against a similar corpus of native speaker writing as well as other
components of the ICLE corpus (French and German L2 writers) in order to ascertain
82
whether features of the Swedish L2 writing are likely to be L1-induced or more
general phenomena of L2 writing54.
Her results can be summarised as follows: In terms of frequency, the category of
modal expressions as a whole is significantly overused by the learners. The overuse
of modal verbs seems to be a universal feature of learner language, as the French
and German learners show the same tendency. One explanation Aijmer offers for
this, which would also explain the universality of this feature, is that a high frequency
of certain modal expressions (e.g. the modal verb WILL) in English is characteristic of
spoken discourse. The fact that learners (as well as novice writers55) often adopt a
more speech-like style in their writing than English native speakers is well
documented in a number of research studies that have drawn on the ICLE data (e.g.
Granger and Tyson 1998; Meunier 2000). Another result from Aijmer‟s study
underpins this impression: The significant and striking overuse of the epistemic
lexical verbs I think and I believe by all learners. I think is regarded as a feature
indicating writer/reader visibility (involvement) and also acts as a parenthetical
discourse frame („sentence builder‟). It is one of the most frequent phrases in spoken
English and therefore a firm fixture of spoken discourse, again pointing towards the
speech-like style of learner writers. Moreover, Petch-Tyson (1998) showed that all
non-native speaker groups used more features of writer/reader visibility than native
speakers and Granger (1998) proves the learners‟ preference for active „discourse
frames‟ over passive ones.
Another feature of learner writing, and possible explanation for the overuse of
modal expressions as a whole, is the tendency to use pleonastic expressions, i.e.
where the same modal meaning is expressed twice in the same utterance by different
means, such as can perhaps, must certainly etc.56 For the Swedish learners, this
general tendency could further be reinforced by interlingual factors. “Whereas in
English, modal meanings are prototypically and frequently expressed by modals, in
Swedish, the epistemic modal meanings are more often realised as an adverb or as
an adverb plus a modal verb” (Aijmer 2000:72). As the modal verbs therefore have a
weaker position in Swedish than in English, the use of pleonastic modal expressions
might well be a factor for the higher frequencies.
Other results from the study showed that the overuse of certain modal expressions
corresponded directly to the overuse of certain modal functions. MUST for example
was overused only as a marker of deontic necessity, not of epistemic. This creates
54
A detailed account of the methodology of learner corpus investigations will be given in chapter 4. 55
See 4.2.2 for a definition of the term „novice writer‟. 56
Cf. also results in Lorenz (1998:64) for the same tendency in the area of adjective intensification.
83
the impression of a very „strong opinion‟ through a direct and emphatic style, which is
not always appropriate to the given topic and certainly contributes to the rhetorical
effect of a text, often marking it as non-native like without overt grammatical or lexical
errors. This uncertainty in striking the right balance between tentativeness and
assertiveness by learners also shows in the overuse of epistemic MIGHT in contexts
where a more strongly assertive modal would make the writer seem less “equivocal,
diffident or naïve” (Hyland and Milton 1997:186).
Aijmer‟s discussion makes it clear that there is no single factor that explains the
diverging patterns of modal expressions between learners and native speakers, but
that a number of interlingual and developmental aspects have to be taken into
consideration as well as the influence of teaching material on learner production. At
the same time, Aijmer (2002:72) admits that
“It is important to point out however, that the study was non-exhaustive and
although it considered a variety of modal expressions the ways of expressing
modality are so numerous that it would require a completely manual study,
using a corpus manually tagged for every kind of modal expression, before any
firm statements could be made about global use of modality in learner writing.”
It is exactly this kind of shortcoming that the present study aims to address.
There are a number of other studies that aim to identify the specific characteristics
of modality use in L2 English using corpus linguistics methods. Hinkel investigates
the effects of different cultural values and norms (Hinkel 1995) as well as different
essay topics (Hinkel 2009) on the use of modal verbs in L2 writing and finds that
“frequency rates of obligation and necessity modals in L2 essays are, in fact, affected
by the writing topic to a great extent. On the other hand [...] the frequency rates of
possibility and ability modals appear to be less topic and context-dependent.”
Similarly, Kwachka and Basham (1990), Basham and Kwachka (1991) and Byrd and
Reid (1998) conclude in their respective studies that in L2 writing, modal verb uses
are affected by contexts and different systems of cultural values and norms, as well
as expectations of discourse and persuasion. Xiaoling and Xiao (2008) carried out a
comparative study of modal verb use between Chinese writers of English as a foreign
language and American native speaker writers. They found that the Chinese
students' writing featured a significantly higher percentage of modal verbs than the
native speakers‟ writing. Furthermore, they uncovered that Chinese students usually
only employ modal verbs in one specific meaning and tend to be blunter and less
tentative than the native speakers. Viana (2006) found that Brazilian learners of
84
English significantly overuse modal verbs, especially WILL, and ascribes this to the
fact that learners “tend to use structures which characterise the oral production of
speakers of English as a first language” (Viana 2006:84). Vethamani, Manaf and
Akbari (2008) investigate Malaysian learners of English and also showed that
learners overuse WILL as well as CAN and COULD to express certainty and ability
and underuse modals expressing probability/possibility.
Although these studies are concerned with L2 English, they will be useful in the
discussion of the findings in this study, in order to determine whether there are
characteristics of modal verb use by language learners that are independent of the
L2, i.e. that occur in learner English as well as learner German. However, it is difficult
to compare the findings of these studies as they use different categorisations for the
modal verbs, sometimes distinguishing between epistemic and non-epistemic uses
and sometimes not. The study by Vethamani, Manaf and Akbari (2008) is also
problematic as it uses a corpus of narrative compositions rather than argumentative
texts. This is not a text type that is very suitable for modal verb use investigations as
the frequencies of modal verbs is relatively low and the functions of modal verbs can
be expected to differ from those in argumentative texts. This hinders further the
comparability of the results.
Additionally, a major drawback of all of the studies in this section is that they,
again, exclusively deal with modal verbs and neglect other modal expressions. Two
studies that seek to avoid this are Gabrielatos and McEnery (2005) and McEnery and
Kifle (2002), who include modal adverbials in their corpus investigations. These
studies are only concerned with epistemic modality but also find differences in the
assertiveness with which learners from different cultural backgrounds present their
arguments. McEnery and Kifle (2002), however, tentatively conclude that these
differences could also be influenced by teaching materials. The relationship between
modal verb frequencies in native speaker corpora and second language teaching
materials is also the topic of a study by Mindt (1996). He finds that the use of
grammatical structures in textbooks for teaching English differs considerably from the
use of these structures in L1 English. WOULD, CAN and WILL were the most
common modals in his research corpus, but in the teaching materials he investigated
WILL is only introduced in the second year of study, whereas the infrequent modals
MUST and MAY are introduced much earlier in the first year. On the basis of these
findings he recommends that frequency of usage should be used as “a guide to
priority for teaching” (Mindt 1996:245f). The influence of teaching material will be
another factor that will be taken into account when discussing the results of the
present study.
85
Lindemann (1996) is the last study that will be discussed in this chapter and is the
only detailed investigation of modality in advanced learners of German to date. Unlike
the previous studies, however, it deals exclusively with modal verbs and disregards
entirely any other modal means. Lindemann (1996:2) stresses in the preface to her
work that, in her view, a study of learner language always has to be concerned with
the individual, actual learner, as language acquisition has to seen as a learner-
individual process. The main focus of her longitudinal study is therefore the detailed,
qualitative and individual analysis of the interlanguage development of every
individual subject with respect to the acquisition of the German modal verb system.
The informants for this study were pupils in the last two years at a Norwegian
secondary school (16-18 years), who were classed as intermediate learners, and
students at a Norwegian university during their first and second year of study, who
had to have passed German at school and were classed as intermediate to advanced
learners. The data collected from all informant groups comprised classroom
observation, informal interviews and written texts and includes ca. 1500 written and
1500 oral occurrences of modal verbs.
Lindemann‟s investigative focus mainly centres on questions of transfer from the
L1, the influence of the learners‟ first foreign language (English), the relationship
between individual learner‟s interlanguage development and the respective learner
group they belong to, and the role of input and material. She bases her analysis on a
contrastive description of the forms and functions of the German and Norwegian
modal verb system and discovers that the fact that several of the German and
Norwegian modal verbs are not only formally congruent (e.g. KÖNNEN/KUNNE) but
also fulfil equivalent functions in both languages (e.g. WOLLEN/VILLE – volition,
wish) plays the most significant role in the development in the learners‟ modal verb
systems.
Despite great variety between individual learners, she distils the following general
trends: The learners‟ modal systems develop continuously from school to first year to
second year of university. At the end of school, the learners have generally only
acquired what Lindemann (1996:27) calls the “Standardrepertoire” (standard
repertoire). This comprises the four modal verbs that are form/function equivalents in
German and Norwegian, i.e. where L1-transfer is positive: KÖNNEN/KUNNE in the
functions of epistemic or deontic possibility, ability and permission; MÜSSEN/MÅTTE
as epistemic or deontic necessity; SOLLEN/SKULLE as request/demand;
WOLLEN/VILLE as volition, plus the L2-specific modal verb DÜRFEN as permission.
Characteristic of the school leavers‟ modal system is furthermore the persistent use
86
of WOLLEN and SOLLEN for future events (instead of WERDEN) or as substitute for
subjunctive form WÜRDE, and the use of NICHT MÜSSEN for prohibitions. These
occurrences can be traced to L1-interferences, as Norwegian forms the future with
the cognates VILLE and SKULLE. In the case of WOLLEN this is probably further
reinforced by the positive transfer in the learners first foreign language – English
(VILLE/WILL for future events). At the end of the first year of university study, the
students have generally expanded their modal verb systems by at least one of the
L2-specific modal verbs (DÜRFEN, MÖGEN, MÖCHTE) although their productive
use is still very limited. A minority of students additionally manage to avoid the
incorrect L1-transfer of WOLLEN, SOLLEN and NICHT MÜSSEN. At the end of the
second year of study, all students have at least partially acquired the L2 specific
modal verbs into their repertoire, but Lindemann (1996:144) notes that even then, “an
interlanguage system that equates to the target-language norm could still not be
attested […] in any of the learners”57.
Apart from the influence of the first language and other foreign languages learnt
before, Lindemann identifies input and teaching materials as a significant factor in the
development of the learner‟s modal verb system. The relative infrequency of certain
modal verbs and modal verb meanings in authentic texts (e.g. MÖGEN) may
contribute to the learners‟ insecurity about how to use these appropriately on top of
the fact that there is no equivalent modal verb in their native language. In this
respect, it is a definite shortcoming of Lindemann‟s study that she does not consider
other strategies of expressing modality.
In this chapter, I have aimed to give an overview of existing studies of modality in
second language research. Each of them contributes considerably to our
understanding of the development of the system of modality in second language
learners and the factors that influence this development. It should, however, also
have become obvious that the previous research on learner modality has left a gap.
There is, as yet, no study of modality in advanced learners of German that takes into
account not only modal verbs but also all the other means that have been identified
as equally valid – and in some cases even more prevalent – markers of modality. It is
the goal of the present study to address this void and fill it with insightful information.
57
“Ein den zielsprachlichen Normen entsprechendes Lernersprachensystem ist allerdings […] bei keinem der Lerner zu verzeichnen.”
87
4. Methodology
4.1 THE CORPUS LINGUISTIC BASIS OF THE STUDY
The methodological foundations of this study lie firmly within corpus-based linguistics.
Countless articles and monographs have laid out, discussed and defended the value
of the corpus approach to linguistics against intuition-based approaches58. I will
therefore not recount the full stretch of the argument here, but rather point out only a
few of the key characteristics of corpus linguistics in general and of learner corpus
research and then concentrate on the situation for German learner corpus research,
as this is the main aim of the present study.
The rapid development of ever more powerful and affordable computer resources
in the past decades has promoted corpus-based studies to the forefront of linguistic
research. But technical advancement alone would not be able to change linguistic
analysis, if it was not grounded in a fundamental change in what linguists perceive as
the essence of language – from “Language as System” to “Language as Instance”
(Halliday 1992)59, focusing strongly on performance data. Therefore, corpus
linguistics as a method for analysing computerised language data is intrinsically
rooted in a Firthian60 view of language that places meaning at the centre of interest,
where meaning configures itself within discourse through lexis and grammar. There is
no sharp distinction between lexis and grammar as both manifest themselves in the
process of creating meaning, as much as there is no distinction between meaning
and pattern61 (Sinclair 1991). Lorenz (1999:1) writes in his excellent introductory
summary of the debate between the two approaches to linguistics:
“Over the last decade and a half, corpus-based linguistics has gained more and
more ground, up to the point of becoming the more popular approach in most
(European) quarters. […] But despite the current popularity of data-based
analysis, it would be wrong to assume that modern corpus linguistics is only
58
The reader is referred here to a small selection of relevant works, e.g. Leech and Svartvik (1975); Aarts and Meijs (1990); Aarts (1991); Sinclair (1991); Leech (1992); Svartvik (1992); Stubbs (1993); Hunston (2002). For German see Lemnitzer and Zinsmeister (2006).
59 This fundamental distinction has been discussed by many linguists under many different terms: „Intuition-based‟ vs. „observation-based grammars‟ – „Chomskyan‟ vs. „Firthian‟ linguistics – „Corpus Linguist vs. Armchair Linguist‟ etc.
60 Cf. the original writings of J.R. Firth (e.g. Firth 1951; 1956). A brief overview of Firthian principles can be found in Stubbs (1993).
61 This means that, firstly, if a word has several senses, each sense will tend to be associated most frequently with a different set of patterns. Secondly, the pattern/meaning association means that words with the same pattern tend to share aspects of meaning.
88
interested in external facts, with a disrespect for theoretical abstraction.”
In fact, the most widely followed analytical procedure in corpus-based studies
seems to aim at synthesising intuitive reasoning and language production data by
incorporating them into a cyclical research routine: Native speaker intuition or
introspection is an efficient means for forming initial hypotheses. On top of this,
already codified knowledge of language, e.g. in dictionaries or grammars, can be
drawn upon, even if it is not corpus-based. However, parts of this knowledge are
inaccessible to introspection, or as Hunston (2002:20) puts it: “Intuition is a poor
guide to at least four aspects of language: collocation, frequency, prosody and
phraseology” (cf. also McEnery and Wilson 1996:12; Alderson 2007). Corpora can
provide the necessary quantitative data to test these initial hypotheses, which, in turn,
is used to refine the original assumptions. When corpora are used in this way of
reconciling theoretical and empirical approaches to language, we usually speak of a
corpus-based approach. The other typical way of using corpora is in the so-called
corpus-driven approach, which takes corpus data as the starting point and aims at
developing linguistic theory through forming hypotheses on the basis of systematic
corpus analysis (cf. Hunston 2002).
The study in this thesis presents a mixture of the two approaches by embracing
the following statement from Hunston (2002:92): “It is important to recognise that no
method of working is neutral with regard to theory. […] Although it may be considered
an ideal to „have the data speak for itself‟, in practice what the data says will depend
to a large extent on how it is able to be accessed.” It might therefore be advisable to
conceive of the corpus-based/corpus-driven division as two ends on a continuum,
rather than either/or categories, where the present study fits in towards the corpus-
based end, but encompasses some corpus-driven elements. As the goal of this study
is to investigate learner modality, in order to first of all establish and delineate the
scope of the study, knowledge from theoretical accounts of modality has to be drawn
upon (cf. chapter 2). Some of this knowledge is corpus-based (e.g. parts of Nehls
1986; Zifonun 1997), some of it is purely introspective (e.g. Brinkmann 1971).
Together, these accounts define which items are investigated in the learner corpus.
The actual analysis of the learner data, however, is more corpus-driven. The aim is to
look at the different modal expressions in order to discover the usage patterns that
are indicative of learner language and extract from them hypotheses about modality
in learners‟ interlanguages and possible implications for the characteristics of learner
language in general. This requires a close scrutiny of the functions that the modal
expressions fulfil in the learner output. This interpretative step is captured in the
89
following statement by Biber, Conrad and Reppen (2004:9): “A crucial part of the
corpus-based approach is going beyond the quantitative patterns to propose
functional interpretations explaining why the patterns exist. As a result, a large
amount of effort in corpus-based studies is devoted to explaining and exemplifying
quantitative patterns.”
4.1.1 Learner corpus research
The methodological footing of the present study implicitly presupposes the validity
and usefulness of learner corpora in the study of learner language. Much the same
as with corpus linguistics in general, there is nowadays a vast range of literature that
convincingly argues exactly this point, especially when the focus is on advanced
learners (cf. Walter and Grommes 2008). Granger (1994) suggests that, in addition to
native corpora, researchers should also compile learner corpora, i.e. electronic
databases of learners‟ interactions in their L2, in order to ensure that their linguistic
needs are accurately and adequately described as well as met. This enables
researchers to ascertain patterns of difficulty in learners‟ foreign language
development, to attend to them through calculated pedagogical intervention, and to
compare systematically learners‟ L2 productions with native speaker productions,
provided they are available in comparable corpora. Again, I will not repeat the full
account of the argument here and instead refer the reader to some key publications,
in particular the pioneering work of Granger and her colleagues at Louvain-la-Neuve,
Belgium, in the development of the International Corpus of Learner English (ICLE),
e.g. Granger (1994; 1998a; 2002), Granger et al. (2002)62. Lorenz (1999) presents a
concise yet compelling rationale for learner corpus research (cf. also Nesselhauf
2004 and Tono 2002). Therein, he traces the development of learner language
research from error analysis and the concept of „interlanguage‟ and positions learner
corpus research within the context of the advantages and criticisms of these two
approaches63. Lorenz concludes with the remark that
“with the recent upsurge in corpus linguistics, and in extending its methods to
the analysis of non-native varieties of English, it is indeed plausible that we
should return to analysing learner language as a separate linguistic variety, only
62
Cf. also publications on the Child Language Data Exchange System (CHILDES), which was first established to investigate child first and second language acquisition, but has since grown to incorporate a wide variety of spoken data, including older second language learners; see e.g. MacWhinney (2000).
63 Cf. also Weinberger (2002) for an overview of the historic development of learner language research.
90
this time with the help of computers, and with the aim of discovering learners‟
deviations that would otherwise go unnoticed.” (Lorenz 1999:8)
Up to now, most of the research effort in learner corpora has focused on English
as a second/foreign language. While the amount of learner corpus resources for
English has grown substantially over the last decade64, other languages are lagging
behind. This certainly does not affect the methodological validity of the corpus
approach to learner language. But it does, of course, impact on the amount of
research output for those languages where the availability of learner corpora is very
restricted. One such language is German. There is a small error-annotated corpus
(95 texts, 27695 words) of German learners with English as L1 that was collected at
Lancaster University (Weinberger 2005), but this is not publically available. Similarly,
an unpublished corpus of L2 telecollaborative data between American students of
German and German students of English was collected at Pennsylvania State
University (Belz 2004). To date, the only publically available German learner corpus
is the Fehlerannotiertes Lernerkorpus (Falko) (error-annotated learner corpus),
collected by Anke Lüdeling and her colleagues at the Humboldt Universität in Berlin65
(e.g. Lüdeling et al. 2005; Lüdeling et al. 2008). This corpus of advanced learners of
German with a variety of L1 backgrounds is continuously being extended and can be
accessed online via a web-interface66. It comprises a so-called „core corpus‟ and
several extension corpora. All subcorpora are part-of-speech tagged; the core corpus
is additionally error-tagged using a multi-layer stand-off annotation model (see
Lüdeling et al. 2005).
The systematic annotation of errors is essential for one of the two main ways in
which hypotheses of second language acquisition are tested with learner corpora.
This is referred to as „computer-aided error analysis‟ (cf. Dagneaux et al. 1998). On
the basis of the error-annotated data, areas of difficulty can be established and it can
be investigated what the potential reasons for these difficulties are. We are not
concerned with this approach here. The study presented in this thesis is based on the
second type of learner corpus research: „contrastive interlanguage analysis‟.
64
For an overview of English learner corpus resources see Nesselhauf (2004) and Pravec (2002). 65
N.B. Falko was not yet available at the time of data collection for the corpus in the present study. 66
For up to date information on corpus size and design and to access the corpus, see http://www.linguistik.hu-berlin.de/institut/professuren/korpuslinguistik/forschung/falko.
91
4.1.2 Contrastive Interlanguage Analysis and the Multiple Comparison Approach
Contrastive interlanguage analysis (CIA) is defined as the quantitative comparison of
learner language productions and native speaker language in order to detect specific
features of learner language (Granger et al. 2002:12f). This approach intrinsically
embraces the notion of learner „interlanguage‟ (IL), introduced by Selinker (1972):
Interlanguage describes the learner's underlying mental knowledge of the second
language at a particular point in the course of learning. The linguistic concepts in this
interlanguage are not only potentially different from the target language, they are also
involved in various dynamic processes and therefore constantly changing. CIA is
therefore comprised of two types of comparisons as depicted in Figure 2:
FIGURE 2: CONTRASTIVE INTERLANGUAGE ANALYSIS (CIA) FRAMEWORK (GRANGER 2002:12)
The first type of comparison in the CIA framework is between non-native speaker
(NNS) data – i.e. interlanguage data from the learner corpus – and native speaker
(NS) data, in order to detect non-native features of learner language, i.e. patterns of
overuse, underuse or misuse. A crucial issue for the validity of this kind of
comparison is the selection of an appropriate native speaker corpus. This selection
essentially defines the „norm‟ against which the leaner data is measured, therefore
factors such as medium, genre and text-type have to be considered, and also the
level of proficiency of the native speakers, i.e. novice writers vs.
accomplished/professional writers, as Lorenz (1999) demonstrates (cf. also 4.2.2
below).
The second type of CIA work involves NNS versus NNS comparisons, i.e.
comparing different learner populations with each other. These learner groups can
CIA
NNS NS NNS NNS vs. vs.
92
differ, for example, with respect to their mother tongue backgrounds, which helps
determine which features of learner language are more likely to be L1-induced and
which are developmental, i.e. shared by several learner populations. Another aspect
that can distinguish learner populations is proficiency. A comparison of learners at
different stages of proficiency can give a more detailed insight into the developmental
patterns of certain learner language features.
The most fruitful approach to studying learner language phenomena with learner
corpora has proven to be one that combines the different types of CIA with yet more
comparative data derived from corpus-based contrastive analyses. This model of
analysing learner corpus data is called “Integrated Contrastive Model” (Granger 1996;
Gilquin 2001) or “Multiple Comparison Approach” (Tono 2002). Figure 3 illustrates the
intricate net of multiple comparisons:
FIGURE 3: MULTIPLE-COMPARISON MODEL FOR LEARNER CORPUS RESEARCH (ADAPTED FROM
TONO 2002A)
The idea in this model is to utilise a variety of corpora to investigate a linguistic
phenomenon from more than one angle. Contrastive interlanguage analysis is carried
out to detect specific patterns in the learners‟ interlanguage – either compared to
native speakers or other sets of learners. Detailed, corpus-based contrastive analysis
Contrastive Interlanguage Analysis
Learner Corpus 1 Learner Corpus 3 Learner Corpus 2
L1 Target Language
Native speaker corpus
Native speaker corpus
Contrastive Analysis
Translation corpus
Contrastive Interlanguage
Analysis
Contrastive Interlanguage
Analysis
93
is carried out on corpora of the first and target language. Although traditional
contrastive analysis, as practiced in the 1960s and 1970s, was largely discredited as
uninformative and inaccurate due to methodological problems (James 1998:4), the
recent efforts of contrastive analysis based on authentic (corpus) data have proven
more reliable. They are usually more modest, making no claims about the
predictability of learner difficulties, and can even challenge some of the claims made
in the intuition-based contrastive analysis literature of the past (cf. e.g. Gilquin
2001:100). The results of this type of contrastive analysis can therefore be used as a
diagnostic tool to establish areas where L1-transfer can explain why the learners‟
production deviates from the native speakers‟. In pulling CIA and contrastive analysis
together we can achieve a triangulation of information that can help not only to
discover specific learner patterns, but also to diagnose their roots. Although the
pedagogical implications of such findings are not the primary concern of this thesis, I
agree with Lorenz (1999:12) on the usefulness of our endeavour for foreign language
teaching:
“It is assumed here that, once patterns of non-native deviance have been
discovered, students can be explicitly made aware of these patterns, and that,
given time, motivation and opportunity to practise, they will eventually be able to
modify their linguistic behaviour into a more native-like direction.”
The next section explains how the multiple comparison approach is applied in the
study of German learner modality presented in this thesis.
94
4.2 INVESTIGATING GERMAN LEARNER MODALITY WITHIN THE
MULTIPLE COMPARISON APPROACH
Figure 4 below replicates the multiple comparison model of Tono (2002a) from the
previous section, but this time indicating the specific corpora that are used in the
present study on German learner modality. The individual corpora will be described in
detail in the following sections.
FIGURE 4: MULTIPLE-COMPARISON MODEL FOR THE STUDY OF MODALITY IN LEARNER GERMAN
4.2.1 The German learner corpus: CLEG
CLEG is the Corpus of Learner German that forms the analytical centre of this study.
It has already been discussed in the introduction that modality occupies a pivotal
place in the construction of arguments. The decision to collect a corpus of
argumentative writing is therefore justified on the basis that this particular text type
demands exactly the types of modal expressions that are investigated here. It should
Contrastive Interlanguage Analysis
CLEG YEAR A CLEG YEAR C CLEG YEAR B
L1 Target
Language
KEDS
LOCNESS
Contrastive Analysis
INTERSECT translation
corpus
Contrastive Interlanguage
Analysis
Contrastive Interlanguage
Analysis
LIMAS
95
therefore provide a rich source of relevant data67. The texts chosen for the corpus
can be classified as “expository-argumentative”. This is defined in a purely
operational way as texts where the task instructions imply “the presentation and
weighing up of arguments, writer‟s criticism or systematic outlines of abstract
concepts” (Lorenz 1999:12). Incidentally, expository-argumentative texts are also the
kind of texts that learners are asked to produce most frequently throughout their
study course at Lancaster University. This means that this collection criterion yielded
the largest amount of reasonably homogeneous texts from all year groups68. All texts
are free compositions, which can be broadly divided into text-based tasks and
essays. Text-based tasks are kritische Zusammenfassung (critical summary) and
kritischer Kommentar (critical commentary). The kritische Zusammenfassung is used
in the first two years of study, where students are asked to first summarise a German
text in their own words and then take a critical stance towards the arguments
presented in the text. The kritischer Kommentar is a task from the final year of study,
where students have to develop their own arguments on a topical question, but they
receive a suitable text as background information. Essays, on the other hand, are
independent argumentative texts on general topics such as violence in the media,
tuition fees, the death penalty etc. (see Appendix A for examples). This distinction is
important to keep in mind when carrying out lexical studies as the lexis in a text-
based composition may be influenced by its source text. As the current study is
concerned with very general vocabulary items, the influence on the results presented
here are assumed to be minimal.
The corpus was compiled over two years following the same guidelines that were
applied to the International Corpus of Learner English (ICLE), as set out in Granger
(1998). This means that the following meta data was collected from each contributing
student:
Age
Sex
Father‟s L1, Mother‟s L1
67
On top of this, there simply is no available learner corpus for L2 German which is controlled for text type, learner L1 and other learner variables AND allows access to the source files, which was crucial for the specific annotation required for the study.
68 The only slight deviance to this occurs in the second year of study (Year B), due to different foci in the course design in that year. Firstly, in preparation for their year abroad, the students have to work on an “intercultural project” where they explore differences in student life and culture between British and foreign students by means of a questionnaire. Secondly, many students take a module on the German press and write a critical analysis of either two German news articles or a German and English news article or a critical summary of the history of the German press after WWII. This means that these two projects (which add up to about 63% of Year B tokens) contain large sections of descriptive writing, although both also contain some elements of argumentative writing. The Year B subcorpus therefore also contains longer but fewer individual texts.
96
Language spoken at home
Years of German tuition in school
Number/length of stays in Germany
Other foreign language in acquisition order
Textbooks used during A-level studies
On the basis of these learner profiles those participants were selected that have
British English as their mother tongue (spoken by both parents), have passed A-level
German before entering the university, which equates to between five and seven
years of German tuition at school, and are not mature students (i.e. they were
between 18 and 19 years of age at the start of their degree).
Texts were collected from those students who had given written consent that their
language productions can be used for the corpus. Texts were taken from all year
groups of undergraduate study at Lancaster University. Most students in the first two
years (Year A and Year B) have spent a few weeks in Germany on vacation or as
part of a school exchange. The students in the final year (Year C) have all spent
between six and twelve months in a German-speaking country as part of their “year
abroad”, which is a compulsory part of the degree scheme for all language majors in
the third year of study. Different levels of proficiency are determined by the external
criteria of “year of study”. In accordance with the guidelines by the DfES (Department
for Education and Skills) achieving an A-level in a Modern Foreign Language is
equivalent to level B1/B2 in the Common European Framework of References
(CEFR) for Languages69. During their university course, students are expected to
work at levels B1/B2 and B2 in Years A and B respectively and at level C1 in Year C,
after the year abroad. For practical and confidentiality reasons, it was not possible to
verify these levels for individual learners or to record marks and degree classes, so
their passing the overall assessment each year and moving into the next year was
considered their qualification for the advancing proficiency levels.
For each text that a student produced, a text profile was created detailing:
Text type (essay, “kritische Zusammenfassung”, “kritischer Kommentar”)
69
The Common European Framework of Reference (CEFR) is developed by the Council of Europe and provides a basis for the mutual recognition of language qualifications, thus facilitating educational and occupational mobility.The CEFR is a document which describes in a comprehensive manner i) the competences necessary for communication, ii) the related knowledge and skills and iii) the situations and domains of communication. The CEFR defines levels of attainment in different aspects of its descriptive scheme with illustrative descriptors scale. On the CEFR scale, levels A1 and A2 are classed as “basic users”, B1 and B2 as “independent users” and C1 and C2 as “proficient users”. For detailed descriptions of these levels see: Council of Europe (2001). Common European Framework of Reference for Languages. Cambridge: Cambridge University Press
97
Topic
Number of words
Timed/untimed (homework or exam)
Available reference materials (grammar books, dictionaries, online resources etc.)
Date of production/year group
Learner and text profiles are stored in a Microsoft ACCESS database, where a
unique text ID is created for each entry, which enables the researcher to link each
text to the relevant text and learner information and create subcorpora according to
different specifications if desired. The text ID is devised in the following format that
gives information on year group and point of collection at a glance:
[a1030_04] means the text was produced by learner „1030‟ in their first year of study
(„a‟) and it is the fourth text produced by that learner („_04‟).
Together, this procedure produced a corpus of the following size and make-up:
Year group Learners Texts Tokens No. of different
topics
Year A 54 181 46 139 12
Year B 38 94 65 865 10
Year C 34 176 87 610 14
Total 126 451 199 614 36
TABLE 8: SIZE AND SUBCORPORA OF THE CORPUS OF LEARNER GERMAN (CLEG)
As the data was collected over the course of two years, some of it is truly longitudinal
(from students in Year A that went on into Year B). Due to the year abroad, no truly
longitudinal data is available from Year B to Year C; the majority of the data is
therefore considered quasi-longitudinal70. The maximum number of texts that an
individual student contributed to the corpus is seven, but no more than four in one
year; on average students contributed four texts71.
In terms of size, this learner corpus – as, in fact, most learner corpora – can, of
course, not compete with the general native speaker corpora of several hundred
million words that are available nowadays. Restrictions in the availability of suitable
data and especially time and research resources prohibited a larger corpus, but, for
the present purpose, the corpus should be large enough to warrant quantitative
inquiries and comparisons, but offers the added advantage of being closely controlled
for many learner and text variables. On top of this, markers of modality are very
70
Texts from a further two years were collected after the initial two years, which gives truly longitudinal data for one complete cohort of students. This data has not been digitised yet and is not incorporated into the study presented here.
71 See also 4.3.3 for a discussion of the problem of dispersion.
98
frequent phenomena, so even a smaller corpus should yield plenty of instances. The
medium size also means that the corpus is manageable enough for qualitative
investigations with the help of manual annotations. The result might therefore be
considered to be more reliable and indicative than those from larger corpora where
strict design criteria have to be sacrificed to achieve a bigger corpus size.
4.2.2 The comparable native speaker corpus: KEDS
In order to follow the multiple comparison approach, a comparable native speaker
corpus to CLEG is required. As no such a corpus existed previously, KEDS (Korpus
von Erörterungen Deutscher Schüler72) was compiled in preparation for this study
alongside CLEG. KEDS contains 62 texts (47 different topics), all by different authors,
totalling 82084 tokens.
The main priority here was to achieve comparability in terms of text type. While
modality is a central feature of argumentative writing, it will almost certainly play a
different or less significant role in other text types such as e.g. descriptive or narrative
texts. Text type is suggested to have a distinctive impact on the use of modality,
which in turn calls for a reference corpus to be comparable in that respect. The type
that matches exactly the type of texts produced by the learners in CLEG is the
Problemerörterung (argumentative essay)73, which is a specified part of the German
secondary school curricula. This format is introduced in year 9 and practiced and
developed in the following years, right up until the Abitur in year 12 (formerly year
13). By selecting texts from the Oberstufe (years 10-13) for the comparable corpus, it
is even possible to almost achieve comparability in terms of age group. The pupils in
years 10-13 are between 16-19 years old. The first year students at Lancaster
University included in the CLEG corpus are 18-19 years old. So, while not entirely the
same, there is at least close proximity and even a slight overlap in age groups.
All texts are collected from open-access internet resources that do not require
membership to a forum or internet community. Some of these are so called
“homework forums”74 where pupils post their essays in order for other pupils to get
ideas (or plagiarise) on a variety of popular topics for argumentative essays,
presentations etc. Other resources include examples from websites that give
guidelines and help for writing (and teaching how to write) argumentative essays75.
72
Corpus of argumentative essays by German pupils. 73
As opposed to e.g. the “literarische Erörterung” (a critical discussion based on a work of literature). 74
http://www.hausarbeiten.de. 75
http://www.teachsam.de; http://www6.digitale-schule-bayern.de/ds.py.
99
As no contact details for the authors of these texts are available, it was not possible
to obtain their consent for contributing to the corpus. It was assumed that the pupils
who posted their work on a freely accessible website, which is clearly intended for
others to take ideas from or even copy their work altogether, would not object to the
texts also being used for research purposes. Although the background information on
authors and text production circumstances is sparser for these texts, all texts
included in KEDS have information on the sex of the author, school year, topic and, in
most cases, grade received. None of the given grades were below a B; where there
is no grade provided it is assumed that only pupils who feel sufficiently confident
about their work (i.e. received a good grade), would post their work on websites for
other pupils to take as successful examples.
Sampling data from pupils in secondary school also provides comparability in
another aspect: For both native speaker pupils and foreign language students these
texts are produced in the context of writing for assessment in an instructional setting.
This specific production setting also marks CLEG and KEDS as corpora of novice
writers. The distinction between novice and expert writers, although often used
especially in research on writing processes (e.g. Ackerman 1991; Smith 2008) and
writing in a second language (e.g. Neff van Aertselaer 2008), is not always clearly
delineated. In this thesis, novice writing is defined as in Scott and Tribble (2006:133)
as “unpublished pieces of writing that have been written in educational or training
settings, (often) for purposes of assessment.” This includes writing in the L1 and a
foreign language. In this respect, the learners in CLEG count as novice writers in both
English and German. In contrast to this, expert writers are defined as those who
regularly produce texts in professional settings for publication, e.g. authors of
newspaper articles and editorials, novels, academic research papers, etc. I realise
that this definition is not without problems, but it is in line with previous research (e.g.
Neff van Aertselaer 2008) and certainly workable for our purposes. This is especially
the case as research has shown convincingly that writing processes are similar for
novice writers, whether they write in their L1 or an L2 and, at the same time, different
from expert writers (see e.g. Raimes 1994; Cumming 1994). Furthermore, Taylor
(1986) and Ringbom (1987) suggest that the problems of less proficient native writers
are comparable to non-native writers‟ problems. In view of these findings comparing
foreign language learner writing with expert native speaker writing seems even more
inadequate. The controlled selection process of CLEG and KEDS participants should
ensure that there is no overlap with a population of expert writers in either of these
corpora.
100
4.2.3 The general native speaker corpus: LIMAS
For reasons of text-type comparability, KEDS is the corpus that is primarily drawn on
in the learner vs. native speaker part of the contrastive interlanguage analysis.
Occasionally, however, reference to a general native speaker corpus is required in
order to determine whether certain patterns of modality use could be features of
novice writing in general, rather than specifically those of foreign language writing.
For these purposes, the LIMAS76 (Linguistik und Maschinelle Sprachbearbeitung)
corpus will be used in this study as a general written NS corpus. LIMAS is similar in
structure to the English LOB and BROWN corpora, aiming at a balanced and
representative cross section of written German. It contains 500 texts (ca. 2000 words
each) of various genres in 33 topical categories (e.g. religion, politics, law,
economics, literature, history, geography, medicine, technology, fiction, arts) which
amount to just over one million running words (1,062,624). One of the drawbacks of
LIMAS, apart form its rather modest size, is its age, as the corpus contains texts that
were produced in 1970 and 1971. It has been shown that in this kind of time span,
the use of English modals has changed (cf. Leech 2003) and it has to be assumed
that there might also have been changes in the German modal system since then.
There are much larger and more recent German corpora available, most notably the
DWDS corpus (Deutsches Wörterbuch der deutschen Sprache des 20. Jahrhunderts
– „Digital Dictionary of the 20th Century German Language‟) or the corpora provided
by the „Institut für deutsche Sprache‟ (IDS) in Mannheim77. Both of these are
accessible online, but do not allow access to or annotation of the source files, which
was necessary for my analysis. The two main drawbacks of LIMAS were therefore
acknowledged but had to be neglected in the present study.
4.2.4 The translation corpus for contrastive analysis: INTERSECT
For reliable contrastive analysis data of German and English modal expressions,
corpus data is used to complement and verify the statements in the more
introspective contrastive accounts of the German and English modal systems. With
corpus data, it can be established which modal expressions are actually used in
similar situations in two languages. This corpus-based contrastive data can be
obtained from parallel corpora. These are either comparable corpora, containing texts
76
http://www.korpora.org/Limas. 77
For a detailed and up-to-date overview of available German corpora see Geyken (2009).
101
in two languages that cover the same content within the same text format, e.g. a
German and English newspaper article on the same topic. The other type of parallel
corpora are translation corpora, containing texts in the source language and their
translations into the target language.
The INTERSECT translation corpus is used for these purposes. INTERSECT was
compiled by Raphael Salkie at the University of Brighton (Salkie 1995) and consists
of translations in both directions of fictional and non-fictional German and English
source texts. The corpus is constructed in the following way:
INTERSECT German English English German
Source texts(ST)/ Target texts(TT)
German ST English TT English ST German TT
No. of tokens 519 540 585 693 205 359 200 379
Fiction 6 texts 5 texts
Non-fiction 21 texts 10 texts
TABLE 9: THE INTERSECT TRANSLATION CORPUS
This type of corpus allows the researcher to obtain information on how certain
expressions in one language are translated into another78, and – more crucially – how
often the different ways of translation occur. In this study, the INTERSECT corpus
was analysed with the parallel concordancer ParaConc79 (cf. Barlow 2002). Using
ParaConc, the source and target texts were aligned and could then be searched for
specific modality items, e.g. modal verbs. The results window then shows the
concordance lines for the search word in the source language, while at the same time
displaying their target language translations in a parallel window. The program also
allows for some basic annotation of the data so that e.g. deontic and epistemic
meanings of modal verbs could be separated. Results from these kinds of
investigation proved that so called „translational equivalents‟ in two languages “rarely
have the same distribution or 100% equivalents in parallel corpora” (Salkie
1997:298). This can help, for instance, to empirically underpin claims about the
prevalence of epistemic modal verbs over periphrastic modal expressions in English,
where German supposedly offers the opposite picture (see 2.3.4). Consider the
following examples from the English source-text/German translations part of the
78
We are aware that the language of translations („translationese‟) harbours its own problems (cf. e.g. Schmied and Schäffler 1996). In the face of the unavailability of other corpora, however, we have to neglect these in the current study.
79 see http://www.athel.com/para.html.
102
INTERSECT corpus. In the first one, the modal verb MUST is translated with its
German cognate MÜSSEN, in the second one a modal infinitive is used instead:
[INTERSECT] To do so, the new peace camp must lead public opinion to a brave reassessment of the national "narrative" and rid it of false myths. Dazu muss das neue Friedenslager die öffentliche Meinung zu einer mutigen Neubewertung der nationalen Geschichtsdeutung und deren Befreiung von falschen Mythen bewegen.
[INTERSECT] A number of issues must be addressed immediately to ensure that the ICC can function effectively in its initial years. Damit der Internationale Strafgerichtshof seinen Aufgaben von Beginn an effektiv nachkommen kann, sind noch eine Reihe von Fragen zu klären.
If these observations are quantified, more reliable statements can be made about
the tendency to use modal verbs or other modal expressions for the expression of
certain modal meanings. Altenberg (1998) has developed a measure called “mutual
correspondence” that mathematically captures the degree of correspondence
between pairs of linguistic (grammatical, semantic or lexical) expressions in parallel
or translation corpora. So, for example, if the English modal verb MUST is always
translated as MÜSSEN in German translations and MÜSSEN is alway translated as
MUST in English translations then the mutual correspondence (MC) between MUST
and MÜSSEN is 100%. The formula for calculating MC values is:
MC = (At + Bt) x 100
(As + Bs)
where At and Bt are the frequencies of the compared items in the translations and As
and Bs are the frequencies of the compared items in the source texts. To take the
actual frequencies of MUST and MÜSSEN in the INTERSECT corpus as an example:
Epistemic MUST occurs 26 times in the English source texts (= As) and is translated
as MÜSSEN in the German translations 21 times (= Bt). MÜSSEN occurs 62 times in
the German source texts (= Bs) and is translated as MUST 48 times in the English
translations (= At). The formula then looks like this:
MC = (48 + 21)
x 100
(26 + 62) The MC for the pairing MUSTMÜSSEN is therefore 78.4. In order to determine
whether this correspondence is stronger in one direction that in the other, the
“translation bias” (TB) can be calculated using the following formula (Merkel 2001:4):
TB = Bt x 100
or TB =
At x 100
As Bs
103
In our example, the TB from English to German is 80.8, the TB from German to
English is 77.4.
Applying these calculations to data from the INTERSECT corpus produces
information that can feed directly into the contrastive analysis between German and
English. Where the learners‟ L1 (English) is concerned, another corpus is
occasionally consulted and that is the LOCNESS (Louvain corpus of native English
essays) corpus80. This 325,000 word corpus was compiled as the native speaker
reference corpus for the ICLE corpora and is therefore similar in design to the KEDS
corpus81. This corpus can be used in order to find evidence for transfer or novice
writer phenomena in the learners‟ writing by triangulation. A certain modal
expression, e.g. Ich denke, that is prominent in CLEG can, for instance, be checked
for its frequency in LOCNESS via translation into English I think. This gives
indications whether a certain linguistic behaviour by the German learners might be
L1-induced.
Having described all the corpora that are utilised in this study, we will now turn our
attention to the actual analysis process. This includes a detailed summary of which
items are being investigated, how they are extracted and annotated and what kinds of
quantitative and qualitative analyses are performed on them.
4.3 ANALYSING MODALITY IN THE LEARNER CORPUS
Two steps had to be taken to prepare the raw corpus data for analysis. The first step
was to identify exactly which expressions of modality to extract from the data. The
second step consisted of annotating the extracted forms according to meaning
categories. Only after this step could different analytical techniques be applied to the
data. All searches and concordance retrievals were carried out with WordSmith Tools
5 (Scott 2008).The following sections detail these three processes.
80
http://www.fltr.ucl.ac.be/fltr/germ/etan/cecl/Cecl-Projects/Icle/locness1.htm. 81
Although one drawback of LOCNESS is that it contains many essays on literary topics, which do not exactly match the text types of ICLE, CLEG or KEDS.
104
4.3.1 Selecting and retrieving modal expressions
Modal expressions were selected according to the theoretical criteria outlined in
section 2.3. In order to make this selection comprehensive, and facilitate automatic
retrieval wherever possible, the different types of modal expressions were
categorised according to Gilquin‟s (2002:187) typology of structures:
Modal verbs form a closed class of lexically-based items. These are: MÜSSEN,
KÖNNEN, SOLLEN, WOLLEN, MÖGEN and DÜRFEN.
In order to retrieve modal verbs in the learner corpus, the corpus was part-of-
speech tagged using the TreeTagger (University of Stuttgart)82 (Schmid 1994).
The tagset includes the categories <VMFIN> (modal verb, finite) and <VMINF>
(modal verb, infinitive). Most modal verbs could thus be retrieved automatically for
annotation through searches for the two categories. However, since learner
language contains errors with respect to spelling and conjugation of modal verbs,
not all instances could be retrieved automatically. Some instances had to be
retrieved by searching for specific word forms (e.g. „*konnen‟). This was possible
since the texts had originally been digitised manually, so a list of common
malformations had already been assembled at that point.
The semi-modal verb WERDEN is classed as an auxiliary verb <VA*> in the
TreeTagger tagset. Therefore, all instances of the lemma WERDEN had to be
checked manually to distinguish modal uses (as defined in 2.3.1.2.7) from others
e.g. passives.
Periphrastic modal expressions are part of a set of open class, non-lexically based
items including nouns, adjectives, adverbs and verbs. This means that the open
classes had to be narrowed down to specific lexical items that could be retrieved
from the data. Zifonun et al. (1997) provide examples in their discussion of these
modal expressions, which were included, as well as additional items from other
publications discussed in sections 2.3.2 and 2.3.3. In order to arrive at a more
comprehensive list of periphrastic modal expressions and because Zifonun et al.
(1997) do not include periphrastic modal expressions for deontic meanings, two
dictionaries were consulted: Dornseiff (2004) and Wehrle and Eggers (1993). Both
of these, rather than listing entries alphabetically, are ordered according to subject
areas. Each entry therefore contains an abundance of words from all parts of
speech belonging to the same subject area or concept. For example, in Dornseiff
(2004) in the chapter Handeln (action), we find the subheading Erfordernis
82
http://www.ims.uni-stuttgart.de/projekte/corplex/TreeTagger/DecisionTreeTagger.html.
105
(necessity), which is concordant with the concept of deontic necessity that can
also be expressed by the modal verb MÜSSEN. In this entry, we find items such
as erfordern, erforderlich, nötig, Notwendigkeit, unabdingbar, unentbehrlich,
unerläßlich etc. (different variants for expressing „necessity‟). With the help of
these dictionaries, an even more comprehensive list of lexical items was compiled
that could be used for automatic retrieval from the corpora. Although
comprehensiveness was the main objective, the categories are so broad that it is
impossible to include all imaginable expressions. Pre-analysis searches
confirmed, however, that many of the items on the original list did not occur at all
in either CLEG or KEDS83. It is therefore fairly safe to assume that the list of 227
items that was eventually entered for the analysis captured the vast majority of
periphrastic modal expressions in the learner and native speaker data (see
Appendix B).
Modal infinitives represent non-lexically based syntactic structures that can be
retrieved from the part-of-speech tagged corpus by searching for the pattern:
(Lemma SEIN or HABEN) co-occurring in the same clause with <PTKZU>
<VVINF> (particle zu (to) immediately followed by an infinitive verb form).
The Konjunktiv mood is either manifest in a specific morphological paradigm on
the main verb (synthetic variant) or, for the Konjunktiv II, in a combination of the
Konjunktiv paradigm on the auxiliary verb WERDEN (i.e. WÜRDE) and an infinitive
main verb (analytical variant). Due to the extensive syncretism between indicative
and Konjunktiv synthetic verb forms, only those instances were analysed where
the Konjunktiv form is morphologically distinct from the indicative. These were
identified by performing word searches that signal the specific usage contexts of
the Konjunktiv as laid out in section 2.3.5.3, e.g. wenn/falls (if) for conditional
clauses, als ob (as if) for counterfactual comparative clauses etc.
All analytical instances of the Konjunktiv II could be retrieved automatically through
a simple word search for WÜRDE. In the same way, all instances of the Konjunktiv
could be retrieved automatically that involve the auxiliary verbs HABEN, SEIN and
WERDEN as all of these have distinct Konjunktiv paradigms. As expected from the
theoretical outlines, these last two types make up the majority of Konjunktiv
occurrences in all investigated corpora.
83
These included, for example, all of the quasi-modal verbs (see 2.3.3).
106
4.3.2 Semantic categorisation and annotation
Once retrieved from the raw data, the modal expressions were then annotated
according to the semantic categories established from the theoretical literature (see
Table 10 and Table 11 below). This process of adding information to the corpus is
what Hunston (2002:79) calls a “category-based methodology, because the parts of
the corpus – the words, or phonological units, or clauses etc. – are placed into
categories and those categories are used as the basis for corpus searches and
statistical manipulations.”
The main binary distinction in this classification is that between epistemic and non-
epistemic modality. Within this distinction, the different meaning categories of modal
expressions are differentiated. As the semantic categorisations applied in this study
are based primarily on Brinkmann‟s (1971) account of the German modal system, the
abbreviations used in the annotations reflect his terminology (see 2.2.2). The
category of epistemic meanings is therefore encoded „I‟, as it denotes conditions for
the assertion of information. Non-epistemic meanings are coded as „R‟, as they
formulate prerequisites for the implementation (German: Realisierung) of an
assertion. The following table (Table 10) gives an overview of the different
subcategories of both „I‟ and „R‟ and the tagset used to annotate the modal
expressions. For ease of reference, the modal verbs corresponding to each category
are also given. The tagset uses abbreviations of the original German classification.
Table 11 gives a translation of the same table.
107
(<Neg> Negation)
(<U> Unklar)
<I> Bedingungen für die Geltung einer Information <R> Voraussetzungen für eine Realisierung
<AZ> Zuversichtliche Annahme, Annahme (oft über zukünftige Ereignisse) aufgrund von Erfahrung
WERDEN
<N> Feststellung, dass Notwendigkeit besteht (Verpflichtung, Zwang)
MÜSSEN
<S> Schlussfolgerung, zwingende Annahme MÜSSEN
<F> Fremde Instanz gibt Vollzugsrichtung vor (Auftrag, Verpflichtung, Erwartung anderer, Versprechen, Gebot, moralische Pflicht)
SOLLEN
<SA>
Abgeschwächte Schlussfolgerung
Eintritt hängt von unbekannten Faktoren ab (Konjunktiv II)
SOLLTE MÜSSTE
<FV> Vorschlag (Vollzug offen) von fremder Instanz SOLLTE
<V> subjektive Vermutung, die dem Sprecher berechtigt erscheint, für die er mit Zustimmung rechnet (Konjunktiv II)
DÜRFTE
<W> Wille, Absicht, Plan, Intention, Bereitschaft
Vollzugsrichtung vom Subjekt aus WOLLEN
<A> Annahme einer Möglichkeit für die Anhaltspunkte bestehen; es ist nichts bekannt, das der Annahme im Weg steht
KÖNNEN MÖGEN
<WA>
abgeschwächter Wille (Wunsch) (aus Horizont hinaus verlagert, daher Konjunktiv II)
MÖCHTE
<AV> Vorsichtige Annahme (Konjunktiv II) KÖNNTE <Z> Zuneigung/Abneigung (negiert) MÖGEN
<BS> Sprecherfremde Behauptung Verantwortung bei grammatischem Subjekt
WOLLEN
<MU> Möglichkeit zum Vollzug gegeben aufgrund äußerer Umstände, erforderliche Mittel, keine Hindernisse
KÖNNEN
<BA> Sprecherfremde Behauptung Verantwortung bei anderen
SOLLEN
<MF> Möglichkeit aufgrund von Fähigkeiten, Eigenschaften, Gewohnheiten
KÖNNEN
<E> Erlaubnis Richtung des Subjekts und Legitimierung durch fremde Instanz
DÜRFEN KÖNNEN
109
<Neg> negation
<U> unclear
<I> conditions for the assertion of an information <R> prerequisites for implementation
<AZ> confident assumption assumption (often referring to future events) based on experience
WERDEN
<N> ascertainment that a necessity exists for a realisation (duty, compulsion etc)
MÜSSEN
<S> conclusion, compelling deduction MÜSSEN
<F>
an external authority dictates the direction of the realisation (instruction, obligation, expectation of others, promise, commandment, moral obligation)
SOLLEN
<SA> tentative conclusion occurrence depends on unkown factors (Konjunktiv II)
SOLLTE MÜSSTE
<FV>
suggestion (realisation uncertain) from an external authority
SOLLTE
<V>
subjective assumption, speculation that seems feasible to speaker, for which they expect agreement from others (Konjunktiv II)
DÜRFTE
<W> volition, intention, plan, willingness subject controls direction of action
WOLLEN
<A> assumption of a possibility for which there are indications; nothing is known that opposes the assumption
KÖNNEN MÖGEN
<WA>
attenuated volition (wish) (outside the subject‟s horizon, hence Konjunktiv II)
MÖCHTE
<AV> tentative assumption (Konjunktiv II) KÖNNTE <Z> affection, liking/aversion, dislike (negated) MÖGEN
<BS> claim made by the grammatical subject, not the speaker
WOLLEN
<MU> possibility of realisation due to external circumstances, necessary means, no obstacles
KÖNNEN
<BA> claim made by others, neither speaker nor grammatical subject
SOLLEN
<MF> possibility of realisation
due to abilities, attributes, qualities, habits KÖNNEN
<E> permission for a realisation direction of subject and legitimisation by external authority
DÜRFEN KÖNNEN
111
As Brinkmann‟s (1971) categorisation was developed on the basis of modal verb
meanings, the other modal means had to be categorised to the same semantic criteria in
order to fit into the modal system. Modal infinitives could be slotted into the deontic
categories necessity <RN>, possibility <RMU>, advice/suggestion <RFV> and
permission <RE>.
For the periphrastic modal expressions, this process was more intricate as each one
had to be matched as closely as possible to one of the existing categories. For epistemic
expressions, this included assessing the degree of certainty that each expression
conveys in an utterance, e.g. sicherlich (certainly) is placed into the category „compelling
conclusion‟. For some items, additional factors such as speaker-internal vs. speaker-
external epistemic source also had to be taken into account. The category <ISA>
„tentative conclusion‟ (modal verbs SOLLTE/MÜSSTE) is roughly on a par in terms of
epistemic certainty with the category <IV> „assumption‟ (modal verb DÜRFTE). The
difference is that <IV> indicates a high degree of „speaker orientation‟, i.e. it is based on
the speaker‟s subjective belief state/attitude towards the proposition (see Brinkmann
1971:391; Mortelmans 1997:211, cf. also section 2.3.1.2.5, whereas <ISA> is neutral in
this respect. Periphrastic modal expressions that explicitly indicate speaker orientation,
such as ich denke, ich glaube, meiner Meinung nach etc. (I think, I believe, in my
opinion) are therefore classed as <IV>, all those that indicate a similar degree of
certainty but no explicit speaker orientation are classed as <ISA>, e.g. wahrscheinlich,
vermutlich (probably, presumably). Although this assignment process was carried out on
the basis of objective criteria as much as possible, some degree of subjective
arbitrariness could not entirely be avoided (see also the discussion of inter-rater reliability
below), as the delineations especially of epistemic modal categories are often more
gradual than clear cut.
Furthermore, it was explained in section 2.3.2.2 that there are periphrastic modal
expressions that do not fit easily into the categories set up for modal verb meanings.
These are, specifically, the assertive modal expressions (encompassing the
subcategories „purely assertive‟, „evaluative assertive‟ and „evidence-focused assertive‟)
and „negative‟ periphrastic modal expressions. These are part of the epistemic
repertoire, but display additional characteristics (e.g. evaluation) that are not captured in
the differentiations of modal verb meanings. Extra categories were therefore set up for
items belonging to these groups. For a full list of categories and items see Appendix B.
112
In a similar vein, Konjunktiv II does not indicate modality in terms of the different
semantic categories that apply to modal verbs, periphrastic modal expressions and
modal infinitives, but rather merely signals the feature [- facutality] in an utterance.
Instances of Konjunktiv are therefore annotated according to their status as Konjunktiv I
and Konjunktiv II and are analysed separately with regard to the different context types
they occur in.
After determining the different meaning categories and converting them into a tagset,
the data was annotated. For modal verbs, modal infinitives and Konjunktivs, this
annotation had to be carried out entirely manually, as a high degree of disambiguation
was required. Periphrastic modal expressions were annotated semi-automatically, using
CoAn84, a German software package for text and content analysis. This program requires
a list of items that can be set either for automatic annotation (e.g. for items like vielleicht
(perhaps), which are unambiguous) or for manual annotation, which is necessary for
items that need disambiguating (e.g. möglich (possible) which can denote epistemic or
deontic modality).
Tags for modal verbs and modal infinitives are constructed from top to bottom level,
simply adding the different codes together. They were inserted as „flat annotations‟ (i.e.
text and annotation are stored within the same file) in the following fashion:
[c1048_04] Wenn wir eine gute Zukunft haben wollen_RW [= deontic volition], so müssen_RN [= deontic necessity] wir uns heute vorbereiten. „If we want_RW to have a good future, we must_RN prepare today.‟
[c2045_01] Letztendlich muss_IS [= epistemic necessity] die Fähigkeit der Gentechnik , Erbkrankheiten und genetische Defizite sehr fruh zu diagnostizieren , als gefährlich gesehen werden. „In the end, the ability to diagnose hereditary diseases and genetic deficits with the help of genetic engineering must_IS be considered dangerous.‟
In clauses containing a negation, the tag is preceded by „Neg‟, regardless of whether the
modal expression or the proposition is negated:
[c2044_06] Hat man das Recht zu entscheiden, sein Leben zu enden, damit er nicht mit unerträglichen Schmerzen leben muss_NegRN, oder...? „Do we have the right to decide to end our lives, so that we do not have to_NegRN live in unbearable pain, or…?‟
84
Developed by Matthias Romppel: http://www.coan.de.
113
In cases where a decision between deontic modality „R‟ and epistemic modality „I‟ could
not be reached due to ambiguity or merger (see 2.3.1.1), the category „U‟ (unclear) is
chosen. This category is most common with the modal verb KÖNNEN (see 5.1.2.4
below):
[c1054_04] Eltern wissen nicht, was ihre Kinder auf dem Internet finden können_U. „Parents do not know, what their children can_U find on the internet.‟
The annotation of periphrastic modal expressions looks slightly different. One of the
constraints of the CoAn program is that it only allows numerical codes as tags. The
different categories from the tagset as well as the additional categories therefore had to
be translated into such numeric codes (see Appendix B for the list of codes)85. CoAn
then inserts these in square brackets after the items to be annotated:
[a2014_03] Wegen dieser Gefahr, vielleicht [240] sollten_NegRFV Eltern ihren Kinder nicht erlauben das Internet zu nutzen. „Because of this danger, maybe [240] parents shouldn‟t_NegRFV allow their children to use the internet.‟
Corrections are inserted for modal verbs and Konjunktivs only. These corrections are
inserted after the tag. For modal verbs a distinction is made whether there is an error in
the selection of modal verb or in verb morphology. If the right modal verb has been
selected in a given context, but its form is wrong, only the correct form is inserted:
[b1011_07] Wir mussen_RN &müssen# von die Nazizeit lernen. „We have to learn from the Nazi era.‟
If the wrong modal verb is used in a given context, the appropriate expression plus the
according tag is inserted:
[c1054_02] Mit dem Verbot muss_RN man auch alle Kruzifixe usw. auch verboten und Nonnen müssen_NegRN &dürfen_NegRE# ihre Ordenstracht nicht im Unterricht tragen. „With this ban one has to also ban all crucifixes and nuns *need not (target: must not) wear their habit when teaching.‟
85
These can easily be converted into the tag format used for the modal verbs. In the present investigation, the numerical codes were kept in order to be able to distinguish modal verbs from periphrastic modal expressions.
114
For Konjunktivs, the annotation of errors signals cases where an indicative form is
applied in cases where Konjunktiv is correct or vice versa. For indicative forms, a
distinction is made between preterite (PR) and present tense (IND) forms:
[b1003_05] Natürlich wenn wir weniger arbeiten_IND &arbeiten würden_KII# , würden_KII die Stressniveaus nehmen ab. „Of course, if we work_IND &worked_KII#86 less, stress levels would go down.‟
[c2028_01] Diese Chemikalien töten Schädlinge und deswegen haben die Räuber kein Fütter. Als Folge von der Benutzung von Schädlingsbekämpfungsmittel hätten_KII &haben_IND# wir die Nahrungskette verändert und natürlich ist das keine gute Sache... „These chemicals kill vermin and therefore the predators have no food. As a consequence of the use of pesticides we would have_KII &have_IND# changed the food chain and that is, of course, a bad thing…‟
[c1052_03] Die heutige Schülen bieten den Kindern an, die Chance sich frei zu außern und ihre eigene Meinungen zu formulieren. Wenn die Schüler so diszipliniert wurden_PR &würden_KII# wie in den früheren Zeiten , würde_KII diese Möglichkeit weggenommen. „Schools today offer the children the chance to speak freely and formulate their own opinions. If pupils were_PR &would be_KII#87 disciplined in the same way as in former times, this possibility would be taken away.
Errors are inserted between „&‟ and „#‟ signs, so the strings between them can be
disregarded in automatic searches and counts, e.g. when using corpus analytical
programs such as WordSmith Tools (Scott 2008). Annotated in this way, they could,
however, also be included in searches without manually adding aberrant spellings to the
search list.
The annotation of errors in the depicted fashion is certainly not without problems.
Lüdeling (2008) discusses these problems in great detail. The main issue that is relevant
here is that a flat annotation, unlike a multi-level architecture, only allows for one tag to
be inserted for each item in question. For error annotation, this means that the
researcher has to decide at the time of annotation how a particular error is categorised
without the option of including alternatives. This categorisation depends on the so called
„target hypothesis‟, i.e. the native speaker‟s interpretation of how an erroneous utterance
86
The annotation of the English translation is for illustration purposes only; the verb form “worked” is, of course, not an English subjunctive form.
87 As before, the English translation is for illustration purposes only, hence the (incorrect) form “would be” in the error tag. “Were” does not reflect the fact that in German, a distinct form (i.e. Konjunktiv II) is required instead of the preterite form. The error emerges probably exactly because, in English, the preterite form “were” is used in this type of conditional clause.
115
should be corrected. Lüdeling (2008:131) has shown that one and the same error can
evoke several different target hyptheses. For example, the following sentence can be
corrected in two ways to resolve the subject-verb agreement error:
[a2027_02] Das ist sehr gefährlich , aber ich denke, daß dieses Problem This is very dangerous, but I think that this problem.SING
leicht gelöst werden können. easily solved become can.PL „This is very dangerous, but I think that this problem can be solved easily.‟
target hypothesis 1: Das ist sehr gefährlich, aber ich denke, daß dieses Problem leicht gelöst werden kann.
target hypothesis 2: Das ist sehr gefährlich, aber ich denke, daß diese Probleme leicht gelöst werden können.
Even this relatively simple error can be interpreted as a number error on either the
subject or the verb. Other errors can prompt many more different target hypotheses that
locate the error in very different places of the utterance and, as a consequence, result in
very different error annotations. A multi-level annotation architecture allows for all of
these alternative hypotheses to be recorded and tagged at the same time, whereas in a
flat annotation, only one target hypothesis can be reflected in the error tag. The choice of
error tag is therefore subject to individual annotators‟ preferences and inter-rater
inconsistencies that can severely limit the indicative power of the error categorisation.
This was, indeed, a problem that arose during the annotation process for the present
study. Its severity, however, was attenuated by the fact that the error annotation was
only limited to expressions of modality, which meant that in cases of doubt, the target
hypothesis was chosen that reflected the error on the modal item in question. In this way,
the annotation of errors involving modal expressions could be kept relatively consistent.
The second problem with error tagging on learner data – inter-rater reliability – was cut
out as there was only one native speaker annotator. This, however, obviously creates a
whole new set of issues. With only one researcher and annotator working on the raw
data, decisions about not just errors, but also about any other forms of annotation and
interpretation are always in danger of being subjectively skewed. These decisions
involve, for example, judgements on what is grammatical or acceptable in the target
language, and these judgements are by no means immune to inter-rater disagreements.
(For in-depth discussions of these issues see e.g. Schütze 1996:77ff; Fetzer 2004:12ff).
Judgements on style or pragmatic force are even more difficult to carry out as a single
researcher. As the nature of the corpus data prevents communication with the authors of
116
texts, it is, for example, difficult to decide whether the learner in the following utterance
used MÜSSEN deliberately in order to convey a strong necessity, or whether they
actually wanted to convey attenuated necessity, but were not able to deploy the correct
subjunctive form MÜSSTE.
[b1040_01] Wenn es wirklich Gleichberechtigung gabe, dann ?muss (target: müsste?) man anerkennen, dass es Hausfrauen und Hausmännern gibt. „If we really had equality, then we have to (would have to)88 acknowledge that housewives as well as househusbands exist.‟
The hypothetical, irrealis context set up by the conditional clause speaks for the
subjunctive form, but the indicative form is by all means conceivable here.
With any of these decisions, a more objective result could be achieved through native
speaker judgement tasks for grammaticality or acceptability involving several test
persons or through a group of annotators, where inter-rater reliability scores can be
calculated. If these scores are high, then the annotation can be considered consistent
and reliable89. Unfortunately, the circumstances of the current study as research for a
PhD thesis did not allow for these measures. As pointed out before, the upside of this is
that the way the data is annotated and interpreted should be internally consistent and
every effort was made to base any decisions on grammaticality, acceptability and
pragmatic force or intentions on objective criteria as much as possible.
4.3.3 Analytical techniques
The analytical process in this study follows a data-driven agenda. This means that, while
the data was prepared on the basis of pre-defined categories of modal meanings, the
analysis focused purely on the way that the data presented itself with respect to these
categories. For this, several analytical techniques were employed within the contrastive
interlanguage analysis framework.
88
Note that the sequences of tenses/subjunctive forms in German conditional clauses is much freer than in English and a mix of indicative and subjunctive forms is not uncommon (see 2.3.5.3)
89 For a discussion of consistency and accuracy in human annotator post-editing of automatically tagged data cf. Baker (1997).
117
4.3.3.1 Quantitative and qualitative investigations
Firstly, quantitative comparisons were conducted for each category of modal meaning
and each type of modal expression. This involved, e.g. frequency counts of individual
modal verbs, which lead to a comparison of the rankings of modal verbs between
different learner groups (Year A to Year C) and between learners and native speakers.
Frequency counts were also conducted in order to establish similarities and differences
in the distribution of types of modal expressions, e.g. the proportion of modal verbs to
periphrastic modal expressions between the learner and native speaker corpora.
Furthermore, frequencies were established within categories and types, e.g. the
distribution of individual periphrastic modal expressions within each meaning category.
Secondly, qualitative investigations were carried out especially in those areas where
the quantitative analysis had uncovered discrepancies between learners and native
speakers. This required close scrutiny of individual concordances, taking into account
the surrounding context of the modal expressions, in order to detect whether there were
any qualitative differences in the use of certain modal expression in addition to
quantitative differences. Aspects that were discovered included e.g. specific collocational
and colligational patterns. This facilitated the detection of areas where the learners‟
usage of modal verbs simply diverged from native speaker patterns but also of specific
problem areas for learners.
These two processes serve to answer the first two research questions: Which modal
expressions do learners use to express which modal meanings at different stages of
proficiency and where and how do these patterns differ from native speakers‟. The third
research question, i.e. identifying reasons for the diverging learner patterns can then be
answered through interpretation of the results in the light of current language acquisition
theories.
4.3.3.2 Interpretation of results
As explained in 4.1.2, the corpus methodological approach to learner language within the
„contrastive interlanguage analysis‟ framework rests on the concept that learner
language can be described in terms of evolving „interlanguages‟. These encompass the
learners‟ underlying mental knowledge of the second language at a particular point in the
course of learning. Interlanguages show linguistic characteristics of the learners‟ L1 and
118
L2, but also characteristics that seem to be independent of L1 and L2 (cf. Selinker 1972).
According to Selinker (ibid), the following processes shape the design and development
of interlanguages:
language transfer – the use of the L1 or other L2s as resource
overgeneralisation – the use of an L2 structure in situations where a different structure
would be used by a native speaker
strategies of language learning – processes that learners use to systematically
generate, test and revise their hypotheses about the L2, e.g. simplification
strategies of second language communication – processes that learners use in order
to achieve successful communication, e.g. formal or functional reduction
transfer of training – the influence of teaching materials, exercises and teaching
progression on language production, e.g. overuse of certain lexical items or
grammatical structures due to overrepresentation in text books
He further identifies „fossilisation‟ as a key characteristic of learner language, where
“fossilized linguistic phenomena are linguistic items, rules and subsystems which
speakers of a particular NL [native language] will tend to keep in their IL [interlanguage]
relative to a particular TL [target language] no matter what the age of the learner or the
amount of explanation and instruction he receives in the TL” (Selinker 1972:215).
All of these aspects are taken into consideration when interpreting the results from the
present study in order to identify reasons for particular learner patterns in the use of
modal expressions. The process here is not to test a specific theory, but to explore which
of the factors that influence interlanguage can serve to explain the data. Where transfer
is concerned, recent theories (e.g. Jarvis and Pavlenko 2008) have identified cross-
linguistic similarity, frequency and prototypicality as crucial factors influencing which
items are prone to transfer. The quantitative results from the various corpora will provide
useful information on all of these aspects. Similarly, overgeneralisations and learning
strategies, such as simplification, can be detected by identifying overuses of certain
structures or items and categorising the semantic functions that learners use them for.
The avoidance of certain items or the replacement of items to fill semantic gaps can also
point to language learning processes. Strategies of second language communication can
be expected to produce features of learner language that are independent of L1 and L2.
The use of fixed phrases or formulaic expressions, for example, constitutes a way of
meeting communicative needs while at the same time reducing the strain on the learner.
Finally, teaching materials must be taken into consideration as influencing factors on the
119
learner output.90 The textbooks considered in this study are “Brennpunkt” (Sandry et al.
2000) and “Zeitgeist”(Herman et al. 2001), which the learner profiling found to be the
most widely used A-level text books that the students in the corpus had used in school.
These also include grammar sections. During their university studies no set textbook was
used, but exercise sheets and materials that the tutors put together for the students from
various sources were also looked at. The recommended grammar for the students was
“Hammer‟s Grammar” (Durrell 2002), and this was also taken into consideration.
4.3.3.3 Statistics
As a large part of the analysis in this study rests on quantitative investigations, it has
to be stressed that modality is conceptualised as a „probabilistic‟ notion here. Modality is
considered an integral part of argumentative writing and, as such, is expected to be
manifest in a variety of forms in the corpora. The interest here is, thus, to determine how
they are realised and with what frequency. Whenever these frequencies are referred to
in comparisons between different corpora (which are of different size), they are therefore
normalised to „frequency per 10,000 words‟. This might be an unusual reference figure,
but as several modal expressions that are reported on are relatively infrequent, it makes
the numerical lists far more readable than higher reference figures of 100,000 words or
even 1 million words.
The statistical significance of differences in frequencies between the learner
subcorpora and the various native speaker corpora as part of the contrastive
interlanguage analysis was tested using log-likelihood statistics (cf. Rayson and Garside
2000) on the raw frequencies. This has been chosen for practical reasons as well as the
fact that other statistical hypothesis testing methods, such as Chi-squared, have been
shown to be less reliable on relatively small corpora and lower frequency items (Dunning
1993; Rayson et al. 2004). Log-likelihood calculations are used to determine whether a
difference between two corpus samples is statistically significant, or in other words,
whether the difference is due to chance. In the present study this operation has been
applied to most frequency counts where learner data is compared to native speaker
90
The textbooks referred to in this chapter are “Brennpunkt” (Sandry et al. 2000) and “Zeitgeist”(Herman et al. 2001), which the learner profiling found to be the most widely used A-level text books that the students in the corpus had used in school. These also include grammar sections. During their university studies, no set textbook was used. The recommended grammar was “Hammer‟s Grammar” (Durrell 2002).
120
data. Log-likelihood values have a direct link with statistical significance: the higher the
value the greater the probability that the difference in the two samples is due to chance.
The following values mark the cut-off points for different degrees of significance:
LL-value ≥ 3.84: The probability of differences being due to chance is p < 0.05
(5%). This marks a significant difference (indicated by *).
LL-value ≥ 6.63 The probability of differences being due to chance is p < 0.01
(1%). This marks a highly significant difference (indicated by **).
LL- value ≥ 10.83 The probability of differences being due to chance is p < 0.001
(0.1%). This marks an extremely significant difference (indicated
by ***)
As all statistical calculations, log-likelihood has its drawbacks. One of the issues in the
specific design of the learner corpus CLEG is the problem of dispersion. This means that
with some items it occurs that a handful of learners use them excessively, thus boosting
their frequency, although many learners do not use them at all (cf. also Lyne 1985).
Another issue is the fact that some learners contributed more texts or longer texts than
others to the corpus. The length of texts ranges from around 100 words to 2500 words,
with a mean of 442 words and a standard deviation of 385. On top of this, learners
contributed from as little as two to as many as seven texts to the corpus, which means
that the interlanguage of some learners is overrepresented. While these problems are
recognised, an undogmatic stance is taken in the present study and all suitable texts
were included. Within the research design for this study, statistical significance is merely
a corroborative argumentative tool. This means that significance tests are taken as
additional evidence for certain tendencies that are detected in the learner data. However,
they never form the basis of decisions to exclude certain categories or expressions form
closer analysis; all categories and modal means are investigated in detail, regardless
whether any statistically significant differences between learner and native speaker
corpora were established.
To conclude, the methodological basis of the present study rests on the supposition
that a learner corpus, such as CLEG, is a valid instrument for uncovering otherwise
undetectable patterns of learner language and that the analytical procedures described
in the multi-comparison model produce valid results. It is, of course, not proposed that
this is the only way of procuring reliable information on learner language, but it should
have become clear that for a study of a linguistic phenomenon such as modality, which is
121
manifest in such a diverse range of lexical and grammatical features, corpus data
provides the most diversely exploitable source of information and the multiple
comparison approach provides valuable results.
122
5. Results
In this chapter, results of the analysis of expressions of modality are presented. Each of
the groups of modal expressions (modal verbs, periphrastic modal expressions, modal
infinitives and the subjunctive mood) is initially dealt with separately. As the modal verbs
in their different meanings set up the semantic categories that this exploration will look
into, they will be dealt with first. The other expressions will then be investigated following
the same semantic categorisations as the modal verbs and the additional categories that
have been identified, e.g. for the periphrastic modal expressions (see 4.3). At the
beginning of each of the sub-sections, general results and observations are given in
order to offer first insights into differences between native speaker and learner usage of
modality. Then, the different types and categories of modality are investigated in more
detail. Generally, there are two (often interlinked) aspects that will be focused on in each
section. On the one hand, comparisons between the learners and native speakers will be
drawn, and on the other hand, comparisons between learner year groups will be made in
order to uncover patterns of progression in the interlanguage system for modality from
less to more advanced learners. Where appropriate, connections between the different
means of modality (e.g. underuse of modal verbs and overuse of periphrastic modal
expressions in the same meaning category) are also pointed out and will be discussed
more comprehensively in chapter 6.
5.1 MODAL VERBS
Overall, 4760 modal verbs were annotated in the learner corpus CLEG and 1388 in the
comparable NS-corpus KEDS. In the general NS-corpus LIMAS, raw frequencies of
modal expressions were retrieved, but no manual annotation was carried out91.
The following table (Table 12) gives raw and normalised frequencies of modal verbs in
CLEG, KEDS and LIMAS.
91
This means that a figure for WERDEN cannot be included from the LIMAS corpus, as WERDEN is used not only as a modal verb but also as an auxiliary verb to form the passive and as a main verb in the sense of "to become" with adjectives and noun phrases, all of which do not carry modal meaning (see 2.3.1.2.7).
123
CLEG (learners) KEDS (comparable NS) LIMAS (general NS)
raw per 10000
words raw
per 10000 words
raw per 10000
words
KÖNNEN 1775 88.9 659 80.3 5413 50.9
MÜSSEN 799 40.0 278 33.9 2565 24.1
SOLLEN 706 35.4 190 23.1 1990 18.7
WERDEN 687 34.4 82 10.0 --- ---
WOLLEN 522 26.2 82 10.0 1161 10.9
MÖGEN 171 8.6 45 5.5 157 1.5
DÜRFEN 103 5.2 52 6.3 326 3.0
TOTAL 4763 238.6 1388 169.1 11612 109.3
TABLE 12: OVERALL FREQUENCIES OF MODAL VERBS IN LEARNER AND NATIVE SPEAKER CORPORA
Two tentative observations can be made from these figures. Firstly, the order of
frequencies of modal verbs in CLEG and KEDS is almost the same (with a slight
variation of MÖGEN and DÜRFEN at the bottom of the table), which in turn is the same
as in LIMAS. This means that in terms of frequency of use in relation to each other (i.e.
for instance KÖNNEN used the most, SOLLEN used more than WOLLEN etc.) the
learners show the same pattern as the native speakers.
Secondly, in general learners seem to be overusing modal verbs in comparison to
both NS corpora. The statistical significance of different frequency scores for the different
items in the respective learner corpora compared to KEDS have been examined through
calculating log-likelihood (LL) values, which are presented in Table 13 (cf. Chapter 5.3.3
for details). Figures preceded by (+) indicate overuse in the learner corpus, (-) indicate
underuse in the learner corpus.
CLEG vs. KEDS CLEG vs. LIMAS KEDS vs. LIMAS
KÖNNEN (+) 4.51* (+) 377.49*** (+) 113.18***
MÜSSEN (+) 4.78* (+) 144.67*** (+) 30.53***
SOLLEN (+) 28.76*** (+) 191.50*** (+) 8.04**
WERDEN (+) 179.37*** --- ---
WOLLEN (+) 79.35*** (+) 240.12*** (-) 0.36
MÖGEN (+) 8.69** (+) 239.47*** (+) 46.15***
DÜRFEN (-) 1.19 (+) 20.56*** (+) 19.70***
TOTAL (+) 137.58*** (+) 1071.19*** (+) 161.37***
TABLE 13: LOG-LIKELIHOOD VALUES FOR DIFFERENCES IN MODAL VERB FREQUENCIES BETWEEN
CLEG, KEDS AND LIMAS
124
(* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
Table 13 reveals that the overuse of modal verbs is statistically highly significant in
both CLEG and KEDS compared to LIMAS (with the exception of WOLLEN for the KEDS
vs. LIMAS comparison, which is very slightly underused in KEDS, but not to a degree
that is statistically significant). This supports the argument that the use of modal
expressions is highly text type dependent and plays a more central role in
argumentative, problem-discussing writing. For this reason, all subsequent analyses of
differences between learner and NS use focus on CLEG vs. KEDS comparisons. Data
from the LIMAS corpus is only drawn upon where the additional dimension of general NS
writing is appropriate and necessary.
However, modal verbs are also significantly overused by learners in comparison to
native speakers writing the same text type (CLEG vs. KEDS). This does not only apply to
the overall figures, but also to every individual year group compared to KEDS, as Figure
5 demonstrates (the differences in frequencies compared to KEDS is statistically
significant for every year group):
FIGURE 5: OVERALL FREQUENCIES OF MODAL VERBS
0.0
50.0
100.0
150.0
200.0
250.0
300.0
Year A Year B Year C KEDS
freq. per
10,000 words
modal verbs
125
Table 13 shows that while this overuse effect is less pronounced (yet still statistically
significant) for KÖNNEN and MÜSSEN, it is highly significant for SOLLEN, WOLLEN and
MÖGEN. DÜRFEN is the only modal that is slightly underused by learners, although not
at a statistically significant level.
In order to see whether there are any changes in these patterns as students progress
from Year A to Year C, Figure 6 presents frequency figures separately for each year
compared to KEDS. Three-dimensional charts such as in Figure 6 are displayed to give
an overview of figures. Corresponding data tables for all such charts are given in
Appendix C.
FIGURE 6: INDIVIDUAL MODAL VERB FREQUENCIES
Figure 6 gives a more detailed picture of where the overall impression of a modal verb
overuse stems from. It can, in fact, be seen that some modal verbs contribute more to
this effect than others:
KÖNNEN is substantially overused in Year A and again quite considerably in Year C,
while actually being underused in Year B as compared to the native speaker corpus92.
92
A possible explanation for the deviances in Year B that result in slightly lower overall frequencies of modal verbs could be the slightly different text types (see Footnote 68 in 4.2.1)
0.0
20.0
40.0
60.0
80.0
100.0
120.0
freq. per
10,000 words
KEDS
YEAR C
YEAR B
YEAR A
KÖNNENMÜSSENSOLLENWERDENWOLLENDÜRFENMÖGEN
126
MÜSSEN is used in almost equal frequencies to the native speakers in Year A and B,
but heavily overused in Year C.
SOLLEN exhibits a similar pattern to KÖNNEN, being overused in Years A and C, but
underused in B.
WERDEN and WOLLEN are both substantially overused throughout all 3 years.
DÜRFEN is actually underused in Years A and B, but slightly overused in Year C.
MÖGEN exhibits exactly the opposite pattern, being overused in Years A and B, but
slightly underused in Year C.
Table 14 replicates the frequency scores from Figure 6 and indicates where the
different frequency scores of modal verbs for the three learner year groups in
comparison to the NS-group in KEDS are statistically significant at p < 0.01. This table
also highlights another interesting fact, namely that the frequency ranking of the modal
verbs for Years A and B are identical, yet considerably different from KEDS, while Year
C is almost identical with KEDS with the exception of SOLLEN and MÜSSEN being
reversed93. This means that, while overall the frequency ranking of modal verbs is the
same as the native speakers‟ for the whole learner group, split up into individual year
groups (and thereby assuming different proficiency levels) some remarkable changes
take place during the progression from Year A to Year C.
Year A Year B Year C KEDS
freq. per 10,000 words
modal verb modal verb freq. per 10,000 words
freq. per 10,000 words
modal verb modal verb freq. per 10,000 words
110.8> KÖNNEN KÖNNEN 64.5< 95.8> KÖNNEN KÖNNEN 80.3
35.5 MÜSSEN MÜSSEN 35.2 53.8> SOLLEN MÜSSEN 33.9
28.6> WOLLEN WOLLEN 23.8> 46.0> MÜSSEN SOLLEN 23.1
27.3 SOLLEN SOLLEN 16.5< 26.5> WOLLEN WOLLEN 10.0
11.1> MÖGEN MÖGEN 12.4> 8.6 DÜRFEN DÜRFEN 6.3
2.6< DÜRFEN DÜRFEN 2.4< 4.3 MÖGEN MÖGEN 5.5
38.4> WERDEN WERDEN 34.2> 32.5> WERDEN WERDEN 10.0
TABLE 14: STATISTICAL SIGNIFICANCE OF DIFFERENCES IN MODAL VERB FREQUENCIES BETWEEN
LEARNER YEAR GROUPS AND KEDS (FIGURES FOLLOWED BY „>‟ INDICATE SIGNIFICANT
OVERUSE IN LEARNER GROUP; „<‟ INDICATES SIGNIFICANT UNDERUSE IN LEARNER GROUP)
93
N.B. the modal verb WERDEN is excluded from this ranking here, as its status is slightly different from the other modal verbs. (see 2.3.1.2.7)
127
In order to find out more about why these patterns of use emerge in the way they do,
it is necessary to look at the different types of modality – epistemic and non-epistemic –
and the individual categories of modality more closely.
5.1.1 Epistemic modal verbs
5.1.1.1 Overview
Lindemann (1996), in her study of Norwegian learners of German, found out that most of
the learners she investigated (after roughly equal times of instruction in the L2) could
only use in their productions what she calls the "Standardrepertoire" (Lindemann 1996:
27) (standard repertoire) which consists of those German modal verbs that are similar in
both form and meaning to those in the L1, plus the L2 specific modal verb DÜRFEN in
the meaning of be allowed to (see 3.2). In order to be able to investigate whether
something like this standard repertoire can also be observed for the British learners of
German, a comparison of the English and German modal verb system is needed. Nehls
(1986:27ff) provides a comprehensive overview of differences and similarities between
epistemic and deontic uses in the two systems, which forms the basis of Table 15
(epistemic modal verbs – this section) and Table 23 (non-epistemic modal verbs, see
section 5.1.2 below). In these tables information from Nehls on which modal verb
cognates in English and German are also equivalent in meaning (gray areas) has been
combined with the classification based on Brinkmann (1971) used for the annotation of
modal verbs in the present study. For example, if a speaker wants to express a
compelling conclusion in German, they can use the modal verb MÜSSEN; if a speaker
wants to express the same kind of compelling conclusion in English, they can use the
cognate modal verb MUST:
(1) Der Dieb muss von den Diamanten gewusst haben.
The thief must have known about the diamonds.
If we want to express a subjective assumption in German, we can use the modal verb
DÜRFTE; in English, however, there is no cognate modal verb that expresses the same
modal meaning, so we have to use a different construction, e.g. probably:
128
(2) Der Dieb dürfte von den Diamanten gewusst haben.
The thief probably knew about the diamonds.
This way, it can be investigated whether learners show any kind of pattern in their modal
verb use that relates to form and meaning equivalences, which in turn would strongly
suggest that transfer plays an important role here.
EPISTEMIC USES OF MODAL VERBS
NECESSITY
Schlussfolgerung MÜSSEN MUST compelling conclusion
abgeschwächte Schlussfolgerung
MÜSSTE OUGHT TO more tentative conclusion
SOLLTE SHOULD
POSSIBILITY
Vermutung DÜRFTE is probably
subjective assumption will probably
Annahme einer Möglichkeit KÖNNEN CAN
assumption of a possibility MÖGEN MAY
vorsichtige Annahme einer Möglichkeit
KÖNNTE COULD
tentative assumption MIGHT
zuversichtliche Annahme (Gegenwartsbezug)
Präsens present tense
confident assumption (present) WERDEN WILL
94
zuversichtliche Annahme (Zukunftsbezug)
Präsens BE GOING TO
confident assumption (future) WERDEN WILL/SHALL
sprecherfremde Behauptung
(des Subjekts) WOLLEN
CLAIM TO PRETEND TO
claim made by grammatical subject
sprecherfremde Behauptung (von anderen)
SOLLEN BE SAID TO
SUPPOSED TO REPORTED TO
claim made by neither speaker nor grammatical subject
(hearsay)
TABLE 15: ENGLISH EQUIVALENTS TO GERMAN MODAL VERBS (EPISTEMIC)
This table gives the impression that for epistemic modality the German and English
modal verb systems are very similar (gray shading indicates modal verbs equivalent in
94
N.B. From a diachronic perspective, the German cognate of English WILL is, of course, WOLLEN. In the synchronic use of both WILL and WERDEN as markers of a possible present or, more often, future event, however, we can assume these two as the form/meaning equivalent pairing (cf. Nehls 1986:25f). The learners have also, by and large, internalised this, as there are only 4 instances of WOLLEN in “confident assumption” contexts in the whole of CLEG.
129
both form and meaning). Therefore, from a predictive contrastive point of view, not many
problems are to be expected. The following chart displays the learner data (in year
groups A to C) and comparable NS data (KEDS) for these categories:
FIGURE 7: EPISTEMIC MODAL VERB FREQUENCIES
From this overview chart it can be seen that, although epistemic modal verbs are
generally less frequent than non-epistemic modal verbs, all different categories (that is to
say, all different modal verbs in their different epistemic uses) are present in the native
speaker data, but quite a few of them are not actively used by all learner groups, namely
SOLLTE, MÜSSTE, DÜRFTE and the two evidentials SOLLEN and WOLLEN. It is
important to note here that some of the modal verbs, and indeed other modal
expressions, only occur witht low overall frequencies. These do not allow for statements
about statistical significance of over- or underuse, but only for those of a binary nature,
i.e. the modal verb in question is either used or not used in the respective learner group.
It also has to be kept in mind that it is problematic to generalise any results from such a
small number of occurrences, which makes their detailed qualitative analysis even more
important. With respect to the low frequency items, a bigger corpus would probably yield
more instances, which would in turn allow for more quantitative evaluations.
0
5
10
15
20
25
30
35
40
freq. per
10,000 words
Year A
Year B
Year C
KEDS
WERDEN
KÖNNEN/KÖNNTE
MÜSSEN
MÖGEN
SOLLTE
MÜSSTE
DÜRFTE
SOLLEN (evid
ent.)
WOLLEN (e
vident.)
130
Another point to make here is this: the fact that these modal verbs were not used by
some learner groups does not mean that these learners do not have passive knowledge
of these meanings/uses or would not, for example, be able to produce them in e.g. an
elicitation exercise. It does mean, however, that they actually did not actively choose to
use them in their free compositions whereas more advanced learner groups and the
native speakers did. It is also remarkable that once a particular modal verb use has
occurred in one year group, from then on it also occurs in the more advanced learner
groups, i.e. there are no modal verb uses that occur, for instance, in Year A and then not
in Year B or C (even though the data is not truly longitudinal).
5.1.1.2 Modal verbs without English form/meaning equivalents: Evidential WOLLEN and SOLLEN, DÜRFTE, MÜSSTE
But what about the predictions from the contrastive analysis? If we compare the
frequency scores for the modal verbs with the list of modal verb equivalents in Table 15,
we can see that those modal verbs that only start being used in higher year groups
match exactly those that have no English form/meaning equivalent modal verb:
Evidential WOLLEN is actually not used at all by any learner, evidential SOLLEN is
attempted twice by a learner in Year C, but only one of these is actually successful:
[c1047_04] Eine von Feshbach entwickelte Hypothese ist die bekannte Katharsisthese, die lautet, dass das Medienerlebnis von Gewalt, zum Beispiel in einem Film, als reinigender Prozess dient, und deshalb ermöglicht das Nachvollziehen der Gewaltakte im fiktionalen Geschehen und nimmt die Zuschauerbereitschaft ab, sich selbst gewalttätig zu benehmen . Diese Katharsisthese soll im Prinzip eine deutliche Aggressivitätsminderung provozieren und hat einen inhibitiven oder hemmenden Effekt. „A hypothesis developed by Feshbach is the well-known catharsis hypothesis, which says that the experience of violence in the media, for example in a film, serves as a cleansing process, and therefore the disposition of the viewers to act violently themselves is being diminished. This catharsis hypothesis is said to provoke a significant decrease in aggressive behaviour and has an inhibiting or restraining effect.‟95
95
The following basic principles are applied in the translations of learner examples: 1. The focus is on conveying the specific point that the example is trying to illustrate (usually in bold print). In the translation of the surrounding context, errors or mistakes that do not pertain to the particular point under discussion are ignored. 2. Only grammatical errors are marked by * and the correct target form is supplied in brackets. Language use that is stylistically questionable is explained in the running text but remains unmarked in the example.
131
The modal verb for subjective assumptions, DÜRFTE, is used once in Year B (where
it is actually within a direct quote) and once in Year C:
[c2032_03] Bundeskanzler Schröder hat zugegeben, dass er zum Teil schuldig ist das die NPD jetzt mehr Einfluß hat. Er dürfte jetzt Angst haben seine Stelle zu verlieren, weil er nicht genug gemacht hat die Rechtsextreme zu bekämpfen. „Chancellor Schröder has admitted, that he is partly to blame for the fact that the NPD has more influence now. He is probably afraid to loose his job now, because he didn‟t do enough to fight the right extremists.‟
MÜSSTE as the modal verb for tentative conclusions is also only used once in Year C:
[c1060_04] Es gab Problemen mit dem Defizit für drei Jahren, die sehr peinlich für die Deutsche sein müsste, als sie aüßerst hart für bestimmte Regeln für Defiziten in Europa gearbeitet hatten. „There were problems with the deficit for three years, which should be very embarrising for the Germans, as they worked very hard for certain regulations concerning deficits in Europe.‟
We can therefore summarise that these modal verbs are only attempted by a few
individual learners in very isolated cases, which means they are definitely not part of a
standard repertoire of modal devices that learners have at their disposal, even in the
most advanced year group.
5.1.1.3 Modal verbs with English form/meaning equivalents: MÜSSEN, KÖNNEN/KÖNNTE, WERDEN, MÖGEN, SOLLTE
On the other hand, the group of modal verbs where English form/meaning equivalent
modal verbs do exist are used throughout all year groups (KÖNNEN/KÖNNTE,
MÜSSEN, MÖGEN, SOLLTE, WERDEN). Within this group, however, we can make one
observation that does not fit the contrastive analysis so neatly. While for the modal verbs
KÖNNEN/KÖNNTE, MÜSSEN and WERDEN the frequency figures generally indicate
either roughly the same use or – sometimes extensive – overuse on the part of the
learners, the modal verbs for the assumption of a possibility (MÖGEN) and tentative
conclusions (SOLLTE) are actually underused by the learners as compared to the native
speakers. SOLLTE is even not used at all by Year A learners.
132
It is at this point that a corpus approach can help improve the contrastive analysis
data in a way to explain this conflict. Nehl‟s synoptic account of differences and
equivalences in the German and English modal verb system offers a summary of what is
linguistically possible, i.e. correct, not on what is likely in terms of frequencies96, that is to
say, how often the respective modal verbs are actually used as equivalents of each
other. In order to obtain this kind of information we can draw upon the translation corpus
INTERSECT described in section 4.2.4 and calculate “mutual correspondences” (MC),
which give a ratio of the frequency with which the respective modal verbs are translated
into each other, for instance how many times epistemic MÜSSEN is translated as MUST
in the German-to-English translations and vice versa. The calculation of “translation bias”
reveals whether this correspondence is stronger in one direction than in the other97.Table
16 shows these three values for the form/meaning equivalent modal verbs:
translation bias
epistemic modal verb equivalents
MC German English
translations English German
translations
MÜSSEN MUST 78.4 77.4 80.8
WERDEN WILL 68.7 62.4 73.4
KÖNNEN CAN 57.8 50.8 82.8
KÖNNTE COULD 51.5 45.5 58.3
SOLLTE SHOULD 26.1 0.0 32.4
MÖGEN MAY 22.0 45.7 8.3
TABLE 16: MUTUAL CORRESPONDENCE VALUES (MC) FOR GERMAN AND ENGLISH EPISTEMIC MODAL
VERB EQUIVALENTS
These figures show high MC values for MÜSSEN and WERDEN and although the
values for KÖNNEN and KÖNNTE are not quite as high, they are still high enough to
argue that the direct modal verb to modal verb translation is still the preferred option in
most cases98. For SOLLTE and MÖGEN, however, the MC values are low, indicating
that the mutual correspondence between the German and English modal verb is not very
96
Although he does include corpus-informed comments on whether certain constructions are frequent or infrequent throughout the detailed discussion; however, no systematic frequency data is given.
97 For a more detailed description of mutual correspondence, translational bias and their mathematical formulas see section 4.2.4.
98 Note also that if the distinction between the categories “assumption of a possibility” and “tentative assumption” is neglected and the frequency counts for KÖNNEN and KÖNNTE and CAN and COULD respectively are combined, this results in an even higher MC value of 60.6.
133
strong. Instead, we find the following examples of how these two modal verbs were
translated in the INTERSECT corpus:
[INTERSECT] Daß es uns aber ein klareres und vor allem gerechteres Bild gibt und daß es ernsthafte Gespräche zwischen den Kulturen erleichtert, das sollte doch niemand bestreiten. However, no-one can deny that it gives us a clearer and, above all, fairer picture which makes it easier to engage in cross-cultural talks.
[INTERSECT] Mag sein, dass der Handel neue Impulse erhält, wenn nach dem EU-Beitritt Polens die Zollkontrolle auf der Neiße-Brücke verschwindet. Perhaps trade and crossborder traffic will receive new impetus when the customs controls cease after Poland‟s accession to the European Union.
Interestingly, the learners follow this pattern exactly – overusing only those German
modal verbs that both share their meaning with the English cognate modal verb and
have a high MC value (MÜSSEN, KÖNNEN, KÖNNTE, WERDEN) and actually
underusing the two modal verbs where a direct modal verb to modal verb translation is
certainly possible but often not the preferred option (SOLLTE and MÖGEN).
5.1.1.4 MÜSSEN (compelling conclusion)
Epistemic MÜSSEN is used in all three learner groups and the normalised figures show
that Year A and B use it less than the native speakers while Year C use it slightly more,
but a closer look reveals that there are also distinct qualitative differences between the
learner groups.
Year A Year B Year C KEDS
MÜSSEN (raw figures) 4 5 16 12
MÜSSEN (per 10,000 words) 0.9 0.8 1.8 1.5
successful instances 2 5 16 12
TABLE 17: EPISTEMIC MÜSSEN
Despite attempting MÜSSEN four times, only two learners in Year A actually do so
successfully. The other attempts are infelicitous:
[a1018_03] Es scheint auch, dass die Kinder, die kein Computer haben, *müssen (target: Ø) verschiedene andere Probleme haben. „It also seems that those children who don‟t have a computer must have several other problems.‟
[a2006_04] Ich liebe Deutschland und es interessiert mich sehr, aber ich kenne keine deutsche Filmschauspieler. Deshalb *müssen (target: dürften) anderen
134
Leute, die Deutschland sich nicht interessiert auch keine Schauspieler kennen. „I love Germany and I‟m very interested in it, but I don‟t know any German actors. Therefore other people, who are not interested in Germany, must also not know any actors.‟
As MÜSSEN expresses a conclusion that excludes any other possibilities, it is
incompatible with a tentative expression like “es scheint” (it seems), with which the first
example is introduced. In the second instance, the speaker cannot have enough
evidence to make that kind of exclusive conclusion. The utterance is beyond the
speaker‟s given horizon, so it would seem more appropriate to use e.g. DÜRFTE as a
subjective assumption or MÜSSTE as a tentative conclusion. Both cases are not based
on transfer errors, as MUST is also inappropriate in the English sentences. It seems,
therefore, that the error lies within the students‟ interlanguage system of German
modality, as the appropriate alternatives (MÜSSTE or DÜRFTE) have not been acquired
yet and this gap is, in these instances, filled with MÜSSEN as a compelling conclusion.
In Years B and C, all instances of epistemic MÜSSEN are correct, for example:
[b1017_05] Es ist of der Fall, dass man Aufstiegsmöglichkeiten in Erwägung zieht, das muss positiv sein. „It is often the case that one considers career opportunities, that must be positive.‟
[c2030_05] Aktive Sterbehilfe oder der medizinisch assistiert Suizid muss als Gefahrlich angesehen werden, und soll verboten sein und verboten blieben. „Active euthanasia or medically assisted suicide must be seen as dangerous and should be and remain forbidden.‟
We can see that while the learners in Year A are still insecure about the use of epistemic
MÜSSEN, in Year B the learners who attempt to use it do so successfully and in Year C
the learners are confident with it and have incorporated it well into their interlanguage
modal system.
5.1.1.5 KÖNNEN and KÖNNTE (assumption of possibility)
KÖNNEN is the modal verb that students in any year group seem to have the least
problems with. Nehls (1986:109) points out that in German KÖNNEN is the default
modal verb to express a possible assumption, while MÖGEN is only available in this
function under certain circumstances (e.g. concessive statements, see 2.3.1.2.6). In
135
English, on the other hand, the default modal verb is MAY, although CAN is also
available for this modal meaning99. There are different constrictions in either modal
system as to when the respective verbs can or must be used, but on the whole German
KÖNNEN is more generically applicable than English CAN100. The same holds true for
English COULD and the German Konjunktiv II form KÖNNTE. L1 English learners
therefore do not need to grammatically or semantically differentiate as much in German
as in English so that they have no problems with the appropriate use of
KÖNNEN/KÖNNTE . This modal verb does, however, pose a problem of a different kind.
As described in section 2.3.1.1, KÖNNEN keeps the Umlaut in the Konjunktiv II form
KÖNNTE, while the preterite form is KONNTE. In English, no such distinction exists as
both forms are represented by COULD. The following table shows that the learners have
difficulties differentiating between preterite and subjunctive form:
Year A Year B Year C
KÖNNEN (raw figures) 70 26 61
KONNTE instead of KÖNNTE 16 8 7
KÖNNTE instead of KONNTE 0 1 2
errors in % 22.9 34.6 13.1
TABLE 18: ERROR RATE FOR EPISTEMIC KÖNNTE/KONNTE
This produces the following types of errors:
[a2019_03] Wenn die deutsche Filmindustrie gleich Erfolg erleben *konnte (target: könnte), wird es fantastich für Deutschland sein, seit diese amerikanische Filme sehr viel Geld verdienen ... „If the German film industry *could experience the same success, it would be fantastic for Germany as these American films make a lot of money …‟
[b1034_01] Ich glaube, dass es sehr gefährlich sein *konnte (target: könnte), wenn man jede in die Zeitung glaubt und … „I think that it *could be very dangerous if you believe everything in the paper and …‟
Some of these errors are perhaps mistakes due to inattentiveness, but the fact that they
also occur in typed texts and that they occur (though less so) the other way round, i.e.
KÖNNTE instead of KONNTE indicates that these are errors due to lack of differentiation
99
Nehls (1986:110) summarises the difference between MAY and CAN: MAY expresses an assumption for a particular single incident whereas CAN expresses a general possibility.
100 Evidence for this can also be seen in the MC scores above (Table 16): CAN is translated as KÖNNEN in
82.8% of cases, whereas only 50.8% of KÖNNEN are translated as CAN. The fact that MÖGEN is much more restricted in its epistemic use than MAY is evident in the fact that only 8.3% of MAY are translated as MÖGEN, but 45.7% of MÖGEN are translated as MAY.
136
rather than “slips of the pen”. We will encounter more evidence that learners have
profound and continuing problems with Umlauts in various forms later on.
5.1.1.6 WERDEN (confident assumption)
Looking back at the overview chart in Figure 7, the one modal verb that stands out
through its extreme and consistent overuse by the learners is WERDEN (the differences
between each year group and KEDS are all statistically highly significant). It is the modal
verb that is used most to express epistemic modality by all learner groups, whereas the
native speakers use a much more balanced amount of WERDEN in comparison to other
modal verbs like KÖNNEN/KÖNNTE. As was explained in section 2.3.1.2.7, WERDEN
can denote present-time or future-time reference, as can the English modal verb WILL.
There are, however, contextual areas where the English use of WILL and the German
use of WERDEN do not match up and they will be the key to exploring why learners
overuse this particular verb so significantly. For this, it is necessary to analyse uses of
WERDEN in more detail and decide for each case whether it has present or future-time
reference.
We will first turn our attention to uses of WERDEN with present-time reference. These
are utterances like the following:
[b1008_03] ...aber ‚Shut endlich up!‟ ist möglicherweise die leidenschaftlichste Verteidigung des Papstes gegen der britische Presse, die man im Internet finden wird. „...but „shut finally up!‟ is possibly the most passionate defence of the pope against the British press that one will encounter on the internet.‟
[keds_055] Trotzdem wird man aber merken, dass ein „Normaler“ im Vergleich zu einem Spitzenathleten nur sehr wenig verdient. „Nevertheless one will notice that compared to a top athlete a “normal” one only earns very little.‟
For present-time reference, the difference between native speakers and learners is
striking:
Year A Year B Year C KEDS
WERDEN (raw figures) with present-time reference
3 8 15 21
all WERDEN (raw figures) 177 225 284 80
as % of all WERDEN 1.7 3.6 5.3 26.3
TABLE 19: WERDEN WITH PRESENT-TIME REFERENCE
137
Table 19 shows that in the NS corpus, over 26% of all sentences containing WERDEN
refer to the present101, whereas the figures for all learner groups barely reach over 5%.
So, while on the whole, learners use WERDEN a lot more than the native speakers, they
hardly ever use it as a „pure‟ modal verb. There are, however, two other (related)
present-time contexts, in which the English WILL is used but WERDEN is not. Nehls
(1986:123) describes these contexts as “habitual WILL” (“charakterisierendes WILL”) on
the one hand and “events of a general nature” (“Ereignisse allgemeiner Art”) on the
other. He gives the following examples for the two:
(3) He‟ll sit there hour after hour looking at the traffic go by.
(4) Don‟t mention it, Kettle. These things will happen.
In both of these contexts the present tense is used in German.
The detailed analysis shows that these uses make up over 10% of all WERDEN in all the
learner groups. (The fact that these uses do not occur in the NS corpus corroborates
Nehls‟ analysis that WERDEN is not used for this context in German.)
Year A Year B Year C KEDS
WERDEN (raw figures) in “habitual” and “general events” contexts
21 23 38 0
all WERDEN (raw figures) 177 225 284 80
as % of all WERDEN 11.9 10.2 13.4 0.0
TABLE 20: WERDEN USED IN “HABITUAL” AND “GENERAL EVENTS” CONTEXTS
Learner examples for these are:
[a2009_03] Heute velassen alle Betrunkner die Kneipe zusammen, und dann werden sie Kampfen schaffen oder Strafrat machen wann sie Banden formieren. „Today all drunks leave the pub together, and then they will start fights or commit crimes when they form gangs.‟
[b1037_01] Es sagt, dass, Frauen werden Arbeitsplätze mit niederer Stellung annehmen weil, sie mehr Biegsamkeit brauchen. „It says that women will take jobs in lower positions because they need more flexibility.‟
While these cases explain some of the overuse of WERDEN by learners, the majority of
it stems from WERDEN with future-time reference:
101
Note that this figure is almost identical to that determined by Pfeffer and Conermann (1977) in their comprehensive corpus analysis. They established that in written language 25% of instances of WERDEN have present-time reference as opposed to 40% in spoken language.
138
Year A Year B Year C KEDS
WERDEN (raw figures) with future-time reference
155 194 229 59
all WERDEN (raw figures) 177 225 284 80
as % of all WERDEN 87.6 86.2 80.6 73.8
TABLE 21: WERDEN WITH FUTURE-TIME REFERENCE
According to Nehls (1986:127), English and German exhibit different patterns in
dealing with future events. WILL and WERDEN respectively can be used to express
future-time events. Theoretically, the present tense can be used in both languages to
express future events if a close link to present time exists (e.g. volition of an agent, event
is already beginning) or if an adverbial marker of future time makes the future-time
reference explicit, although in English this requires the present progressive.
(5) Susanne wird morgen einen Vortrag halten. „Susanne will give a talk tomorrow.‟
(6) Susanne hält morgen einen Vortrag. „Susanne is giving a talk tomorrow.‟
While this is straight-forward in German, because morpho-syntactically there is only one
present tense and one future tense, in English there are more options that compete in
these contexts: There is the GOING TO-future as another possibility of expressing
events that have a close link to present time or that will happen in the very near future
and then there is the distinction between simple present tense and present progressive.
Nehls (1986:128) states that the GOING TO construction is by far the most generally
applicable option in this context, while the use of the present tense is much more
restricted. Leech and Svartvik (1975:72) explain that “the present progressive is used for
future events resulting from a present plan, programme, or arrangement […] the simple
present is used (but not too often) to refer to future events which are seen as absolutely
certain […] In these sentences, we may say that the speaker treats the event as a fact,
and puts aside the doubt one normally feels about the future”. From these remarks we
can summarise that while in German the present tense can be considered a general and
unmarked option for expressing these types of future events102, in English the GOING
TO construction would be the equivalent unmarked option rather than the present tense.
102
Cf. Nehls (1986:127): “… die Annahme über einen zukünftigen Sachverhalt [wird] durch WERDEN + Infinitiv ausgedrückt, während das Präsens in der Regel auf zukünftige Ereignisse hinweist, die sich bereits
139
It therefore seems feasible to argue that in these contexts the learners avoid the
present tense in German in their argumentative essays as, from an L1 English
perspective, they would seem to make claims of absolute certainty that the described
events will happen, while the point of their „pro and con‟ argumentation is that they
explore different possible scenarios. This is why they seek to introduce an explicit
expression of futurity/modality into their sentences in order to allow for „the doubt one
normally feels about the future‟, and in German this explicit expression is WERDEN.
The following concordances exemplify this. In the three pairs of examples similar
propositions on various topics are put forward by the learners and the native speakers,
but while the learners use WERDEN in their statements, the native speakers feel that the
unmarked present tense is sufficient:
[c2043_01] Der Artikel handelt von der Polemik der Gentechnik und, ob die Menschheit wirklich davon profitieren werden oder es wird eigentlich nur problematisch. „The article is about the problem of genetic engineering and whether human kind will really benefit from it or whether it will actually only be problematic.‟
[keds_20] Ich verstehe die Ängste, die bei diesem Thema vorhanden sind, da es sehr neu und noch relativ unerforscht ist, doch muss man vor allem auch die Chance sehen, die uns diese Technik bietet. „I understand the fears that are present with this topic, because it is very new and relatively unexplored, but one has to, above all, see the benefits that this technology offers.‟
[a2013_03] In der Universität und in der Arbeitsplätzen werden das Internet wesentlich für die Übermittlung sein. Es wird besser für die Kindern sein, wenn sie über das Internet lernen, wenn sie jünger sind, weil es einfacher sein wird. „In universities and in the workplace the internet will be crucial for transmissions. It will be better for the children, if they learn about the internet when they are young, because it will be easier.‟
[keds_006] Dies mag daran liegen, dass ihre Eltern bereitwillig ein solches Gerät finanizieren, weil sie wissen, dass die Arbeitswelt ohne Computer nicht mehr denkbar ist und sie ihre Söhne und Töchter optimal auf die Zukunft vorbereiten wollen. „This may be due to the fact that parents readily finance such a device, because they know that a workplace without computers is now unimaginable and that they want to prepare their sons and daughters as well as possible for the future.‟
in der Gegenwart ankündigen.“ (… the assumption of a future event is expressed by WERDEN + infinitive, while the present tense usually refers to future events that have a close link to the present.)
140
[c1049_04] Die Mehrheit der Bevölkerung können zwischen Realität und Fiktion unterscheiden. Die gewalttätigen Szenen werden fast keine Auswirkung haben und es macht man nicht agressiver. „The majority of people can distinguish between reality and fiction. The violent scenes will have almost no effect and it doesn‟t make one more aggressive.‟
[keds_006] Manche besorgten Erwachsenen gehen davon aus, dass das Spielen am Computer den Spieler direkt beeinflusst, also sein Handeln in der realen Welt bestimmt […] Jürgen Fritz legt jedoch in seinem Artikel „Fördern Computerspiele die Gewaltbereitschaft?“ dar, dass brutale Spiele die Aggressionsbereitschaft kaum beeinflussen. „Some concerned adults assume that playing on the computer directly influences the player, so that it determines his actions in the real world […] Jürgen Fritz, however, argues in his article “Do computer games encourage a disposition for violence?”, that violent games hardly influence the propensity for aggression at all.‟
These examples should also make it clear that we are not dealing with errors here, but
rather with an issue at the intersection between pragmatics and grammar. There are
instances in the learners‟ texts where WERDEN is involved in actual grammatical errors,
for example:
[a2025_01] Der erste sehr gross Pop Band der 20. Jahrhundert war die Beatles. Die Mädchen *werden wild auf der Beatles sein. (target: Die Mädchen waren wild auf die Beatles.) „The first really big pop band of the 20th century were the beatles. The girls *will be crazy for the beatles.‟
But these cases are rare. In the majority of cases, the use of WERDEN rather gives the
impression of redundancy, which is both a rather subjective and less quantifiable notion,
as it involves concepts like personal preference and writing style. One factor that
contributes to this impression is the fact that there are more instances of WERDEN as
modal verb/future tense marker in combination with WERDEN or SEIN as the infinite
verb in the learner groups compared to native speakers:
[a1004_02] Es wird ein schwieriger Kampf sein. „It will be a difficult struggle.‟
[b1005_06] Es wird sehr schwierig sein, genug Information nur mit geschlossene Fragen herauszufinden. „It will be very difficult to find out enough information just with closed questions.‟
141
[c1047_01] Es bleibt zu sehen, ob die Todesstrafe bald häufiger als Urteil verkündigt werden wird. „It remains to be seen, whether capital punishment will soon be passed as a sentence more often.‟
English WILL cannot function as a main verb, the infinitve BE is therefore mandatory.
German WERDEN, however, can function as a main verb. In the above sentences this
would be perfectly correct without altering the meaning, which is why the combinations
with the infinitives of WERDEN or SEIN appear redundant.
Another factor is the use of WERDEN for signposting purposes in order to announce
the way a learner intends to structure the essay. These are more prominent in longer
texts and therefore appear a lot in the Year B texts (see Footnote 68 in 4.2.1), although
this type of explicit text structuring is generally employed a lot more by all learner groups
compared to the native speaker group:
[b1031_01] In diesem Essay werde ich die Rolle der Allierten nach dem zweiten Weltkrieg im Wiederaufbau der deutschen Presse untersuchen. Dann werde ich über den Einfluss sprechen, dass die Allierten auf die moderne deutsche presse gehabt haben. „In this essay I will investigate the role of the allied forces in rebuilding the German press after the second world war. Then I will talk about the influence that the allied forces have had on the German press.‟
Yet another factor is that learners sometimes use WERDEN in a subordinate clause
where the main verb already denotes a volitional act or intention (which gives the link to
present time that in German often triggers the use of the present tense):
[c2040_01] Der Präsident des NRW-Verfassungsschutzes möchtet ein neues Gesetz einführen, dies wird die NPD in Deutschland verbieten. „The president of the NRW office for the protection of the constitution wants to introduce a new law that will ban the NPD in Germany.
[b1014_06] ..., weil ich habe beabsichtigte, dass ich mit Leute sprechen werden, die ich nicht gut kenne. „…, because I intended, that I will speak with people that I don‟t know very well.‟
In example [b1014_06], the surrounding context is even a past one, in which case
WÜRDE would be more consistent than WERDEN (as WOULD would be in the English
sentence).
142
Finally, sometimes learners‟ statements sound quite forceful through the use of
WERDEN, because they are made within a wider hypothetical context (subjunctive on
the surrounding verbs):
[c2040_05] Es gibt auch Forschung, die sagt, dass durch Euthanasie zu viele Leute in Europa sterben werden. Letzendlich ist es klar , dass die Einführung eines Gesetzes für Euthanasie schwere folgen für Europa mitbringen würde. „There are also studies that say that through euthanasia too many people will die in Europe. At the end of the day it is clear, that the introduction of a law for euthanasia would bring with it severe consequences for Europe.‟
[c1046_01] Wenn wir verstehen könnten, wiso Leute Morde begehen, könnten wir das Problem an der Basis handhaben. Wenige Leute wollen Straftaten machen - sie sind auf verschiedene Weisen dazu gezwungen. Sie sind auch Opfer und brauchen Hilfe. Langfristig wird es wenigere Kriminellen geben. „If we could understand why people commit murder, we could address the problem at its basis. Few people want to commit crimes – they are being forced in different ways. They are also victims and need help. In the long run there will be fewer criminals.‟
One could therefore argue that the use of the subjunctive form WÜRDE or KÖNNTE
might be more appropriate as it would be consistent with the hypothetical context that the
learners set up. It has to be stressed, however, that this is an issue of writing style and
any comment on the appropriateness of these stylistic choices is, of course, a matter of
subjective preference. There is no way of knowing whether the learners for some reason
failed to follow through with the subjunctive forms throughout the section or whether they
consciously chose WERDEN in order to add force to their argument. It will, however, be
shown in the section on subjunctives (5.4) that the learners generally have difficulties
distinguishing the different morphological forms of WERDEN, i.e. preterite (wurde),
Konjunktiv I (werde) and Konjunktiv II (würde), which would underpin the suggestion that
we are, in fact, dealing with morphological errors here. It can also be asserted that native
speakers tend to use WERDEN less often within hypothetical arguments and rather use
subjunctive forms throughout.
When looking at the development of the use of WERDEN throughout the learner
groups, it has to be noted that all of these „deviations‟ are present in all learner year
groups. A further indicator of this „non-progression‟ is the fact that the differences in
frequencies of the modal verb WERDEN between the three learner groups are not
statistically significant, i.e. there is no significant reduction in the overuse of this verb
143
towards a native speaker level. It can be concluded that, with regard to WERDEN as a
modal verb, the learners seem to be influenced by their L1, not so much on a lexical
level, but rather on a higher order, conceptual level which appears to be quite resistant to
change as the students advance.
5.1.1.7 MÖGEN (assumption of possibility)
As mentioned in section 5.1.1.5 MÖGEN can be used in certain contexts as an
alternative to KÖNNEN to express a possible assumption, but its use is typically
restricted to concessive contexts103 (cf. Diewald 2004: 243). All in all, epistemic MÖGEN
is only attempted 5 times in the learner groups, but none of these are in concessive
contexts. This does not mean that they are incorrect, but to the native speaker these
sentences sound infelicitous, exactly because MÖGEN evokes expectations of a
concessive context:
Year A Year B Year C KEDS
MÖGEN (per 10,000 words) 0.2 0.3 0.2 1.3
MÖGEN (raw figures) 1 2 2 11
concessive contexts 0 0 0 9
TABLE 22: EPISTEMIC MÖGEN IN CONCESSIVE CONTEXTS
There are two felicitous instances where the learners use MÖGEN in semi-fixed
constructions:
[a2018_03] Viele haben auch Angst vor den verlängerten Öffnungszeiten. Es mag sein, dass es zur mehr Gewalt führen würden. „Many people are also afraid of longer opening hours. It may be that this would lead to more violence.‟
[c1046_05] Übrigens ist die Situation gar nicht so schlecht, wie als beim ersten Blick erscheinen mag. „By the way, the situation is not nearly as bad as it may seem at first sight.‟
In the other three instances, MÖGEN is either redundant altogether [b1034_01] or the
context would make KÖNNEN more appropriate [b1011_05; c2036_03]. In these
instances, L1 transfer might have played a role:
103
An analysis in the LIMAS corpus showed that in 100 out of 128 instances (78%) of epistemic MÖGEN occurred in an explicitly concessive context.
144
[b1034_01] Ich *mochte (target: Ø) zu Anfang denke, wie konnte man die oben gestellte Frage beantworten ? „In the beginning I *might think, how could you answer the above question?‟
[b1011_05] Auf der anderen Seit mag (better: kann) ein Mensch Schul haben und so sie müssen jede job machen, um Geld zu bekommen. „On the other hand a person may have debts and so they have to do any job in order to get money.‟
[c2036_03] Die ständige Bombardierung mit Bildern und Musik mögen (better: kann) unerfreulich sein und könnte Ältere möglicherweise leicht stören, auch wenn die Worte eine mächtige Botschaft übermitteln. „The constant bombardment with pictures and music may be unpleasant and could perhaps disturb older people, even if the words convey a powerful message.‟
5.1.1.8 SOLLTE (tentative conclusion)
Epistemic SOLLTE is not used at all in Year A and only once in Year B. In Year C,
however, we encounter 8 cases, 7 of which are correct (0.8 per 10,000 words). This is
proportionally still only about half the amount of occurrences as in the native speaker
corpus (14 instances; 1.7 per 10,000 words), but those learners who attempt it generally
show that they have grasped the use of epistemic SOLLTE, e.g.:
[c2038_01] Schon mit der Erfahrung von BSE sollte man gelernt haben, dass Menschen in solchen grundlegenden Bereichen der Natur sich nicht einmischen sollen. „From the experience with BSE we should have learnt that people ought not to interfere with such fundamental areas of biology.‟
There are, however, some instances where learners fail to realise that only the
subjunctive form SOLLTE is available epistemically and use SOLLEN instead:
[c2042_02] Wenn wir Studiengebühren haben, konnten Studenten je nach Einkommen zahlen […] Wir *sollen (target: sollten) auch nur hoch motiviert Studenten haben, die ihren akademischer Grad benutzen wollen. „If we had tuition fees, students could pay according to their income […] We *shall also have only highly motivated students who want to use their academic degree.‟
While these errors are not frequent in the epistemic category (as overall frequencies are
so low), we will see that this inability to discriminate between SOLLEN and SOLLTE
creates a bigger problem for the non-epistemic modal verb uses.
145
To summarise the category of epistemic modal verbs, we can conclude that there is a
marked difference between those modal verbs where transfer of cognate modal verbs in
German and English is positive and those where this is not the case. The modal verbs
where there are no form/meaning equivalents in German and English (MÜSSTE,
DÜRFTE, evidential WOLLEN and SOLLEN) are hardly used at all and although some of
them start being used in the higher year groups they cannot be considered part of the
learners‟ established system of modality. In the group of modal verb meanings that have
form/meaning equivalents in German and English, we have seen that there is a
difference between those verbs that are equivalent in form/meaning and have high
mutual correspondence scores – MÜSSEN, KÖNNEN/KÖNNTE, WERDEN – and those
that are theoretically equivalent but have low mutual correspondence scores – MÖGEN
and SOLLTE. The modal verbs in the former group are well established (or in the case of
MÜSSEN, become established in the course of Year A) in the learners‟ productive
knowledge and form part of the standard repertoire, while the ones in the latter group are
used much more sporadically and cannot be considered well established. We need to
keep these findings in mind when looking at the results for modal adverbials, as this will
reveal some interesting correspondences. Another factor that was discovered as
influencing the learners‟ modal system is the ability (or rather inability) to distinguish
indicative and subjunctive forms, which in turn has bearings on the use of certain modal
verb meanings.
5.1.2 Non-epistemic modal verbs
5.1.2.1 Overview
For the analysis of non-epistemic modal verbs, a similar approach is employed as for the
epistemic modal verbs. This means, we will first of all look at the corresponding
summarised table of equivalences in the German and English modal verb system:
146
NON-EPISTEMIC USES OF MODAL VERBS
NECESSITY
Verpflichtung
MÜSSEN MUST duty, compulsion
BRAUCHEN NEED/HAVE (GOT) TO (SHALL
104) SOLLEN
instruction, obligation, expectation
Vorschlag
SOLLTE SHOULD
suggestion MÜSSTE
OUGHT TO
SUPPOSED TO
Wille WOLLEN
WANT TO intention, plan, volition,
willingness INTEND TO
WISH
abgeschwächter Wille (Wunsch)
MÖCHTE WOULD/SHOULD
LIKE TO attenuated volition (wish)
Zuneigung MÖGEN LIKE affection, liking
POSSIBILITY
Möglichkeit (Umstände) KÖNNEN KÖNNTE
CAN possibility (circumstance)
COULD
Fähigkeit KÖNNEN KÖNNTE
CAN COULD
possibility (abilities, attributes,
qualities, habits) BE ABLE TO
Erlaubnis
KÖNNEN CAN
permission DÜRFEN
MAY BE ALLOWED
TABLE 23: ENGLISH EQUIVALENTS TO GERMAN MODAL VERBS (NON-EPISTEMIC)
Table 23 indicates that for non-epistemic modality there are actually fewer one-to-one
form/meaning equivalents of German and English modal verbs. However, the ones that
do exist cover a large area of non-epistemic meanings: The basic expressions for non-
epistemic necessity (MUST), suggestion (SHOULD) and possibility (CAN/COULD) have
German cognates with the same meanings – MÜSSEN, SOLLTE, KÖNNEN/KÖNNTE.
The mutual correspondence values (see 4.2.4)for these pairings are as follows:
104
N.B. Nehls (1986) includes SHALL in his overview under the caveat that its use in this sense is very restricted and adds later that in most contexts SHALL and SOLLEN do not constitute meaning equivalents (Nehls 1986:52).
147
translation bias
non-epistemic modal verb equivalents
MC German English
translations English German
translations
MÜSSEN MUST 75.9 72.6 85.5
KÖNNEN/ CAN/ KÖNNTE COULD
62.6 59.5 73.3
SOLLTE SHOULD 56.8 57.0 56.6
TABLE 24: MUTUAL CORRESPONDENCE VALUES (MC) FOR GERMAN AND ENGLISH NON-EPISTEMIC
MODAL VERB EQUIVALENTS
Table 24 shows that all three pairs of form/meaning equivalent modal verbs have high
mutual correspondence values, which means they are available as translational
equivalents in most contexts. Figure 8 below shows whether there is any interrelation
between the information in this table and actual frequency figures:
FIGURE 8: NON-EPISTEMIC MODAL VERB FREQUENCIES
Overall, the frequencies for non-epistemic modal verbs are much higher than for
epistemic ones. All non-epistemic modal verbs are present in all learner groups and,
generally, there seems to be a tendency for overuse of these verbs on the part of the
learners. This obviously means that even in Year A, the learners are beyond the stage of
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
50.0
freq. per
10,000 words
Year A
Year B
Year C
KEDS
MÜSSEN/MÜSSTE
KÖNNEN/KÖNNTE (c
ircum.)
KÖNNEN/KÖNNTE (a
bility)
SOLLTE
SOLLEN
WOLLEN
MÖCHTE
MÖGEN
DÜRFEN
KÖNNEN (perm
iss.)
=
148
acquiring modal verb means for non-epistemic modality, which is to be expected at this
level.
With regards to the German-English form/meaning equivalents, however, the picture
is not straightforward. Comparing the information in Table 23 to the usage patterns of
modal verbs by learners and native speakers reveals that there are hardly any
correspondences.
Out of the three equivalent modal verbs MÜSSEN is the only one that is consistently
(but only significantly in Year C) overused in all three learner groups as compared to the
native speaker group. SOLLTE is significantly overused in Year C, but actually
underused in Years A and B. KÖNNEN can be used in three different non-epistemic
contexts, all of which also allow CAN in English: KÖNNEN as circumstantial possibility is
overused in Year A, but underused in Years B and C, whereas KÖNNEN as ability is
overused in Years A and C, but underused in Year B. Finally, KÖNNEN as permission
occurs roughly equally to KEDS in Year A, but is overused in Years B and C.
But unlike the epistemic modal verbs, where non-equivalence corresponded with
underuse, most non-epistemic modal verbs are overused even if there is no cognate
English modal verb with an equivalent meaning. SOLLEN and especially WOLLEN are
consistently and heavily overused by learners, MÖCHTE and MÖGEN are overused in
Years A and B, while in Year C the frequency drops to roughly the same level as KEDS.
DÜRFEN, the alternative to KÖNNEN for expressing permission, is underused in Years
A and B, but overused in Year C. Again, it is necessary to look at the individual modal
verbs in more detail to understand how the learners use them differently to the native
speakers.
5.1.2.2 MÜSSEN (duty/compulsion) and MÜSSTE (suggestion)
The modal verb MÜSSEN in its non-epistemic meaning expresses necessity, obligation
or compulsion and is equivalent to English MUST and HAVE (GOT) TO105. MÜSSEN is
overused slightly in all learner groups, but there do not seem to be any difficulties for the
learners. The Konjunktiv II form MÜSSTE expresses a mitigated obligation, which gives
105
For English, linguists have not been able to reach agreement over the functional distinction between MUST, HAVE TO and HAVE GOT TO. The most generic description is summarised by Jarvis (1972:246): “If the speaker wishes to mark or emphasise that it is he, the speaker, who makes the proposition necessary, he uses must [..] In all other cases in the non-past must and have to are interchangeable.” In German, MÜSSEN is used for all of these cases, so the learners, in effect, have to apply fewer semantic distinctions than in English, which should make this modal verb unproblematic for the learners.
149
statements with MÜSSTE a notion of advice or suggestions. MÜSSTE is therefore very
close in meaning to SOLLTE and both are equivalent to English SHOULD (cf. Nehls
1986:58), as can be seen in this example:
[keds_045] Damit meine ich, dass einige Lehrer sensibler auf die Art eingehen sollten, mit der ihre Schüler an eine Aufgabe heran gehen und sie bei der Ausarbeitung beobachten und gegebenenfalls korrigieren sollten. Dem Anfertigen einer Arbeit müsste also mehr Aufmerksamkeit gewidmet werden, als dem bloßen Vortrag oder der abgegebenen Endform. „What I want to say is that some teachers should be more sensitive to the way their pupils address a task and they should observe them and if necessary correct them while they are carrying out the task. More focus should be on the process of working on a task rather then the final presentation or the end result that is handed in.‟
While MÜSSEN appears to be well established in the learner‟s repertoire already in
Year A (including its past tense form MUSSTE) and learners generally use it in the same
way as native speakers, MÜSSTE only really starts being used in Year C, as the
following table shows:
Year A Year B Year C KEDS
MÜSSEN (MUSSTE) (raw figures) 123 (36) 173 (51) 344 (9) 220 (9)
MÜSSTE (raw figures) 0 2 25 30
MÜSSTE as % of all MÜSSEN 0.0 0.9 6.6 11.6
TABLE 25: NON-EPISTEMIC INDICATIVE MÜSSEN (MUSSTE) AND KONJUNKTIV II MÜSSTE
Several observations can be made from these figures. First of all, if we look at the figures
for the Konjunktiv II form MÜSSTE, it becomes very clear that there is a distinct
difference between Years A and B on the one hand and Year C on the other. It is only in
this final year that the students are able to semantically distinguish between MÜSSEN
and MÜSSTE and use the subjunctive form with some confidence, although still only
about half as often in contexts expressing non-epistemic necessity as the native
speakers (6.6% vs. 11.6% of all forms of MÜSSEN):
[c1048_01] Das heißt, so ein Verbot wäre nicht gegen der Bundesverfassung. Jedes Land müsste allerdings einen Gesetzesentwurf dafür verabschieden. Bisher haben zwei Länder solche Gesetze verabschieden; „This means that such a ban would not be against the federal constitution. Every federal state would, however, have to pass a separate law for that. So far, two federal states have passed such a law;‟
150
It has to be noted, however, that there is a qualitative difference between the learners‟
use of MÜSSTE and the native speakers. As pointed out before, in German MÜSSTE is
not only used in hypothetical or indirect speech contexts (e.g. conditional clauses), but
can be used alongside SOLLTE to express advice or recommendation and it is in this
function that MÜSSTE is used in 28 out of 30 instances in KEDS:
[keds_017] In diesem Fall nimmt man der Wirtschaft etwas ab , was eigentlich sie tragen müsste. „In this case, the economy is relieved of something that it should actually pay for.‟
[keds_32] Bei Auffälligkeiten bestimmter Hunde müsste konsequent vom Staat durchgegriffen werden - unabhängig von der Rasse des Hundes. Der Halter sollte zur Verantwortung gezogen werden. „If certain dogs behave in a particular way, the state should consistently take drastic measures – independent of the breed of dog. The owner should be held responsible.‟
In the Year C data, however, only 5 out of 25 instances of MÜSSTE are used in this
meaning, 4 of which by the same learner in the formulaic phrase Zusätzlich/Trotzdem
müsste man sich fragen… (On top of that/Nevertheless we would have to ask
ourselves…), which leaves only one other genuine use of MÜSSTE as advice or
suggestion:
[c2044_03] Hinzu kommt die Probleme Einführung der Hartz IV Arbeitsmarktreform und das Gefühl dass die Schere zwischen Arm und Reich immer stärker auseinander geht. Die Regierung müssten jetzt gemeinsam gegen den Nazis arbeiten, aber nicht mit einem Verbot , sondern … „On top of that there are the problems with the introduction of the Hartz IV reform of the job market and the feeling that the gap between rich and poor is getting wider and wider. The government should now work together against the Nazis, but not with a ban but…‟
In all other instances, MÜSSTE, is used in the meaning of obligation, but in a wider
hypothetical context. It therefore seems that the appearance of this particular modal verb
is closely linked to an overall improved grasp of the Konjunktiv II. The following example
shows how the student sets up and manages to maintain a hypothetical context by using
subjunctive forms on modal and other verbs in order to explore various possible
outcomes of changes to tuition fees in German universities:
[c2037_02] Wenn man später bezahlen könnte, würden die Studenten vielleicht unter Druck stehen, weil sie gute Noten und auch eine gute Stelle bekommen müssten, um die Studiengebühren zurückzuzahlen. Deshalb
151
findet man einen Teufelskreis. Wenn die Studenten selbst bezahlen müssten, würden sie lieber nicht in der Universität studieren. „If you could pay later, the students would perhaps be under pressure, because they would have to get good marks and a good job in order to pay back the tuition fees. We therefore find a vicious circle here. If the students had to pay for themselves, they would rather not study at the university.‟
In contrast to this, learners in Year A are often not able yet to use the subjunctive
appropriately in order to explore hypothetical arguments and rather use indicative forms
for both modal verbs and other verbs, as the following example shows:
[a2025_03] Es ist auch kein weg um Wahl zu winnen zum Beispiel, wenn die Steuern steigt. 24 Stunden Öffnungszeiten können die Betrunkenheit steigen und also gibt es mehr unsozial Benehmen. Die britische Kultur *musste (target: müsste) sich daran 24 Stunden trinken gewöhnen um der Gesetz zu funktionieren. Es kann auch mehr Vandalismus sein, weil es mehr Betrunkenheit ist und es gibt mehr Geld aus dem Steuerzahler genimmt. „It is also not the way to win elections, for example, if taxes are rising. 24-hour opening times can increase drunkenness and therefore there is more anti-social behaviour. The British culture *had to (target: would have to) get used to 24-hour drinking for the law to work. There can also be more vandalism, if there is more drunkenness and that takes more money away from the tax payer.‟
Note that the learner in this example might have attempted to use a subjunctive in “Die
britische Kultur musste sich daran…” but confuses the past tense form musste with the
correct form müsste. This confusion over morphological forms of modal verbs regularly
causes errors. MÜSSTE does occur in Year A and B, but in all but two instances in Year
B it is actually used incorrectly.
Equally, there are occasions where MÜSSEN (or MUSSTE) is used instead of
MÜSSTE in contexts where it is not just a matter of choice between stating a necessity
or a mitigated necessity, but where its use constitutes a grammatical error (see example
[a2025_03] above). Finally, the most frequent type of errors constitutes uses of a form of
*MUSSEN as in the following example:
[b1042_02] Die Firmen *mussen (target: müssen) mehr Flexibiliäat zu Arbeitnehmer anbieten oder viel mehr Frauen werden vielleicht ihre Arbeitsplätze einfach verlassen. „The companies *have to offer employees more flexibility or many more women will perhaps simply leave their jobs.‟
152
As with KÖNNEN above, one could argue that missing Umlauts could simply be a
mistake caused by inattentiveness (a „slip of the pen‟) in handwritten texts – which is
probably true for some of them. The fact that these errors occur also in typed texts,
however, points towards conceptual errors. It seems that the confusion over
morphological forms here is paired with and mutually dependent on the inability to
discriminate semantically between indicative MÜSSEN, past tense MUSSTE and
subjunctive MÜSSTE. We will see in the next sections that a similar problem can be
observed with the other modal verb that in its indicative form expresses non-epistemic
necessity and in its subjunctive form recommendation or suggestion – SOLLEN – as well
as the modal verb for non-epistemic possibility - KÖNNEN.
Finally, straightforward L1-interference can be observed in errors involving negation.
In English, negating the modal verb MUST leads to the negation of permission, as in:
(7) You must take your shoes off inside.
(8) You must not wear shoes inside.
In German, the negation of the modal verb MÜSSEN leads to the negation of obligation
(which can also be expressed by NICHT BRAUCHEN, see section 2.3.1.2.1), which in
English is expressed by NEED NOT:
(9) Du musst im Haus deine Schuhe ausziehen. „You must take your shoes off inside.‟
(10) Du musst im Haus keine Schuhe tragen. „You don‟t need to wear shoes inside.‟
In order to express negation of permission (prohibition), the modal verb DÜRFEN has to
be used:
(11) Du darfst im Haus keine Schuhe tragen. „You must not wear shoes inside.‟
Learners in all year groups use NICHT MÜSSEN instead of NICHT DÜRFEN in order
to express prohibitions in some cases, while using the correct form in others. But again
there is a marked difference between Year A and B on the one side and C on the other,
in that in Year C the proportion of correctly used NICHT DÜRFEN is much higher than in
the earlier year groups, although the error rate is still at 17.3%.
153
Year A Year B Year C
*NICHT MÜSSEN as prohibition errors (raw figures)
4 6 8
correctly used NICHT DÜRFEN for prohibition (raw figures)
6 6 38
errors in % 40.0 50.0 17.3
TABLE 26: NON-EPISTEMIC NICHT MÜSSEN AS PROHIBITION
The alternative to negated NICHT MÜSSEN as negated obligation – NICHT
BRAUCHEN – occurs four times in KEDS. It is only attempted seven times in total in all
three learner groups ( Year A: 3; Year B: 1; Year C: 3). Out of these, five instances are
correct [a2026_01], two instances are erroneously used in positive statements
[a2020_03]:
[a2026_01] Wir brauchen nicht andere Sprachen zu lernen wir haben Englisch. „We don‟t need to learn other languages, we have English.‟
[a2020_03] Warum *braucht (target: muss) jeder eine Hochschulausbildung haben?106 „Why does everybody have to have a university degree?‟
5.1.2.3 SOLLEN (instruction/obligation) and SOLLTE (suggestion/advice)
The second modal verb in the category „non-epistemic necessity‟ is SOLLEN. In terms of
bindingness or strength of obligation, it is on a par with MÜSSEN, but SOLLEN indicates
that there is an external authoritative source that directs the obligation imposed on the
subject. The Konjunktiv II form SOLLTE, on the other hand, indicates a suggestion or
recommendation. In the learner data, SOLLEN – although strictly speaking the non-
equivalent – is overused in all groups, while SOLLTE (which is equivalent to English
SHOULD) is only overused in Year C.
As suggested before, the reason for this pattern seems to be a serious confusion of
forms on the part of the learners. In all three learner groups SOLLEN is used extensively
for suggestions, where SOLLTE is correct as the examples below demonstrate:
106
The difficulty here is that BRAUCHEN as a main verb with an accusative complement is a common phrase in German. BRAUCHEN as a semi-modal verb (governing an infinitive) is only used in negative utterances.
154
Year A Year B Year C
*SOLLEN used as suggestion (target: SOLLTE) (raw figures) 54 28 108
all SOLLEN (raw figures) 68 65 232
errors in % 79.4 43.1 46.6
TABLE 27: ERRORS NON-EPISTEMIC SOLLEN AS SUGGESTION
[a2026_02] Sie *sollen (target: sollten) nicht zu viel trinken, hauptsächlich für ihre Gesund aber auch wenn sie oft mit anderer Leute kämpfen, *sollen (target: sollten) sie nicht trinken. „They *shall107 not drink too much, mainly for their health but also if they often fight with other people they *shall not drink.
[b1034_01] Zum Schluss bin ich der Meinung, dass es harte gesetzte geben *soll (target: sollte), weil die Presse in ein sehr mächtiges Position ist. „Finally, I think that there *shall be tough laws, because the press is in a very powerful position.‟
[c1046_02] Das Tragen eines Kopftuches ist eine Ausübung, was niemandem schadet. Wir *sollen (target: sollten) diese Frauen in Ruhe lassen. „Wearing a headscarf is an act that doesn‟t hurt anyone. We *shall leave these women in peace.‟
So, although the learners seem to improve after Year A in their ability to use the right
form SOLLTE for suggestions, the error rate even in Year C is still almost 47%.
Curiously, the difference between SOLLEN and SOLLTE is, in principle, the same as for
English SHALL and SHOULD. This, however, does not seem to be salient enough for
the learners to incorporate this distinction between indicative and Konjunktiv II into their
L2 modal system. An added complication is probably that the preterite form of SOLLEN
matches that of the Konjunktiv II – SOLLTE108.
As the error rate is so high in this category, in order to see overuse/underuse patterns
as compared to the native speaker group it is useful in this instance to consider error-
corrected frequency figures for SOLLEN and SOLLTE:
Year A Year B Year C KEDS
SOLLEN (per 10,000 words) 2.8 (14.7) 5.6 (9.9) 14.2 (26.5) 5.4
SOLLTE (per 10,000 words) 24.1 (12.6) 10.0 (5.8) 37.3 (25.0) 15.1
107
As mentioned before (see footnote 104), SOLLEN is often better translated as IS TO or BE SUPPOSED TO, because SHALL in the sense of an obligation is nowadays usually restricted to legal contexts. It is used merely to demonstrate that the use of SOLLEN here is incorrect.
108 Interestingly, there are no instances of *SÖLLTE in the learner data, this means that learners do not
overgeneralise from MÜSSEN/MÜSSTE and KÖNNEN/KÖNNTE, but the fact that MÜSSEN and KÖNNEN change their vowel in the preterite, but not in the subjunctive and SOLLEN retains the vowel probably contributes to the confusion over morphological forms with these verbs.
155
TABLE 28: ERROR-CORRECTED FREQUENCIES OF NON-EPISTEMIC SOLLEN AND SOLLTE (ORIGINAL
FREQUENCIES IN BRACKETS)
From the error corrected figures we can see that what seemed to be an overuse of the
modal verb SOLLEN in all three learner year groups actually turns out to be an overuse
just in Year C. SOLLTE originally appeared overused only in Year C, but the corrected
figures indicate a significant overuse also in Year A. In order to understand where the
different usage patterns stem from, a closer look at the concordances should reveal
whether there are specific uses or contexts that the learners favour.
For SOLLEN it can be established that there is, again, a qualitative difference, namely
with respect to sentence type, between Years A and B on the one hand, and Year C and
KEDS on the other. In Year C as well as in KEDS, roughly 40% of the instances of
SOLLEN occur in direct or indirect questions, which are often employed as text
structuring devices, either to reiterate the essay question (e.g. at the beginning of the
text or towards the end for the introduction of the conclusion) or as rhetorical questions
(e.g. to introduce a new aspect or angle into the discussion):
[keds_011] Eine wichtige Frage bei der Legalisierung ist, wie der legale Handel geregelt werden soll. Dabei ist unter anderem zu beachten, wie man Jugendliche möglichst effektiv vom Cannabiskonsum ausschließen kann [...] Es muß auch darauf geachtet werden, daß keine zusätzlichen Anbieter von harten Drogen geschaffen werden. „An important question when discussing the legalisation is how dealing is to be regulated. Amongst other things we have to consider how young people can be excluded effectively from taking cannabis […] We also have to make sure that no additional dealers of hard drugs are created.‟
[c2038_03] Im Text wird es erklärt, dass die rechtsextreme NPD mehre Stimmen bekommt. Politiker streiten aber miteinander darüber, warum die Partei wachst und was man dagegen machen kann. Soll man eine solche Partei verbieten? „The text explains that the right-wing NPD is getting more votes. Politicians argue why the party is growing and what can be done against it. Should a party like this be banned?‟
In Years A and B, this kind of use of SOLLEN hardly occurs at all:
156
Year A Year B Year C KEDS
SOLLEN in questions (raw figures) 2 1 30 17
all SOLLEN109
(raw figures) 13 38 126 44
SOLLEN in questions as % 14.3 2.6 37.8 38.6
TABLE 29: NON-EPISTEMIC SOLLEN IN QUESTIONS
In this respect the learners in Year C, despite overusing the modal verb, employ
SOLLEN in a native speaker-like way with regard to the specific text type of
argumentative essays.
Another point that shows the development of the learners‟ use of modal verbs in the
wider context of writing for a particular text type is the distribution of grammatical subject
types. This is particularly relevant for utterances involving SOLLTE, as with this modal
verb the writer gives their assessment of how the controversy or problem that is at the
centre of an argumentative essay should be solved or, crucially, who should take up
action to solve it – the subject of an active sentence. The following chart displays the
distribution of different types of grammatical subjects and passive constructions as
percentages of the overall number of instances of SOLLTE in the respective learner year
groups and native speaker group:
FIGURE 9: TYPES OF GRAMMATICAL SUBJECT/PASSIVE CONSTRUCTIONS WITH NON-EPISTEMIC SOLLTE
109
Error-corrected figures are used in this table.
0%
20%
40%
60%
80%
100%
Year A Year B Year C KEDS
"wir"
animate agent subject
inanimate subject/"es"
"man"
passive
157
Figure 9 indicates that there is a marked difference in the way that the responsibility for
action in order to resolve a problem is assigned by the native speakers and the learners
in the different year groups. The first three categories from the bottom (passive, “man”,
inanimate subject/“es”) indicate that the writer expresses a general suggestion that the
proposition carried forward in the utterance be implemented, but avoids assigning the
responsibility to any particular person or group that could act as an implementing agent.
These are passive sentences (where, typically, the agent remains unnamed), sentences
with generic “man” (one) and sentences that have an inanimate subject or dummy-
subject constructions with “es” (it):
[keds_039] Deshalb sollte bei der Erziehung in der Schule angesetzt werden. Der Religionsunterricht sollte noch ethischer ausgelegt werden... ‟A start should therefore be made with school education. Religious education should be even more oriented towards ethics…‟
[keds_46] Experten halten einen derartigen Schritt schon lange für notwendig und schlagen als neues aktives Wahlalter 16 Jahre vor, das passive sollte man von 21 auf 18 Jahre senken. ‟Experts have been considering such a step as necessary for a long time and suggest 16 as the new active voting age, and one should reduce the passive voting age from 21 to 18 years.‟
[keds_43] Dennoch sollte es jedem Schüler generell offen stehen, neben der Schule zu jobben. ‟It should nevertheless be down to the pupil to decide whether they want to have an after-school job.‟
The graphs in Figure 9 show that in the native speaker data, well over 60% of
SOLLTE is used in these types of sentences that leave the implementing agent
unspecified. In the learner data, there is a distinct difference between Year A on the one
hand and Years B and C on the other. In Year A, barely 20% of instances of SOLLTE
are used with unspecified agents. In Years B and C, this figure reaches just over 40% -
still only roughly two-thirds of the native speaker figures. We can also see that passive
constructions feature little in Years A and B, whereas in Year C the figures match those
of the native speakers. Constructions with “man” are underused in all learner groups.
This means that in the majority of cases in the learner data, SOLLTE is used with
specified agents as subjects that are identified as the person or group that is to take the
proposed action (including the subsection of “wir” (we) as subject):
[a2006_03] Vor Kindern das Internet nutzen erlauben sollen, sollten Eltern Regeln festlegen, Anleitung und zeitliche Begrenzungen geben und elterlicher Regler benutzen.
158
‟Before children are allowed to use the internet, parents should lay down rules, give advice and time restrictions and use parental control.‟
[b1017_06] Ebenso gibt es eine Diskussion ob Politiker eine Maut in Berlin einführen sollte, um die Zahl der Fahrzeuge in der Inner-Stadt zu reduzieren. ‟There is also a discussion whether politicians should introduce a congestion charge in Berlin in order to reduce the number of vehicles in the inner city.‟
[c2029_02] Die Regierung sollte irgendwie eine Lösung finden, um Studienplätze für alle Studenten zu finanzieren. ‟The government should somehow find a solution to finance university places for all students.‟
So, while the native speakers prefer to phrase recommendations and suggestions
with SOLLTE in a way that the responsible agent remains unspecified, the learners often
explicitly specify this agent, although it can be seen that there is a marked development
within the year groups, e.g. in that the use of “man” constructions rises considerably in
Year B and passive constructions become more widely employed in Year C.
5.1.2.4 KÖNNEN and KÖNNTE (possibility and ability)
Non-epistemic KÖNNEN is by far the most frequent modal verb in all the subcorpora. As
with epistemic KÖNNEN, it is also quite unproblematic for the learners. Nehls (1986:106)
points out that “the German modal verb KÖNNEN shares nearly all its usage variations
(including the tendency to be ambiguous110) with English CAN.”111 The modal verb
KÖNNEN therefore poses few conceptual difficulties for learners, as KÖNNEN can be
used in (nearly) all contexts in which English CAN is used and vice versa. Unlike with
modal verbs like MÜSSEN and SOLLEN, with KÖNNEN, the Konjunktiv II form KÖNNTE
is also generally equivalent to the English subjunctive form COULD – another factor that
reduces problems with this modal verb.
Three functions are distinguished in the annotation of non-epistmic KÖNNEN:
Possibility due to external circumstances (example [c2037_04]), possibility due to
personal abilities (example [c2031_02]) and permission. The first two are very close to
each other in meaning (and sometimes not clearly distinguishable from the context) and
110
This is evident in the fact that in terms of frequencies the annotation category “undecided” (between epistemic and non-epistemic modality) is only relevant for KÖNNEN.
111 “… das deutsche Modalverb KÖNNEN nahezu alle Verwendungsweisen mit dem englischen CAN gemeinsam hat (inklusive der Tendenz zur Ambiguität)“.
159
set apart from KÖNNEN as permission. We shall look at these two categories together
first:
[c2037_04] Im Internet kann man die Themen leicht sehen und man kann auch die relevanten Artikel wählen, anstatt der ganzen Zeitung lesen. ‟On the internet you can easily see the topics and you can chose the relevant articles rather than reading the whole paper.‟
[c2031_02] Wenn Leute gut ausgebildet sind, können sie ihre Jobs besser ausführen, und deshalb wird die Wirtschaft von diesen Akademikern geholfen. „If people are trained well, they can fulfil their jobs better and the economy therefore benefits from these academics.‟
Distributions of indicative KÖNNEN versus Konjunktiv II KÖNNTE are about the same in
all three learner subcorpora and in KEDS, i.e. learners use the Konjunktiv II form in
roughly the same proportion as native speakers:
Year A Year B Year C KEDS
indictive KÖNNEN (as % of all KÖNNEN) 92.1 85.6 87.6 88.2
Konjunktiv II KÖNNTE (as % of all KÖNNEN) 7.9 14.4 12.4 11.8
TABLE 30: PERCENTAGE DISTRIBUTION OF INDICATIVE AND KONJUNKTIV II FORMS OF KÖNNEN
The learner data shows some slight over- and underuse of these two meaning variants in
the different learner groups (cf. Figure 8), but generally KÖNNEN and KÖNNTE are both
well established in the learners‟ modal repertoire and the learners use this modal verb in
the same way as native speakers. The only problem that arises with KÖNNTE is the
same that we have seen with MÜSSTE and SOLLTE, namely the inability to distinguish
indicative preterite and subjunctive forms, evident in errors where learners use KONNTE
instead of KÖNNTE and vice versa. While these errors do not seem to be frequent if
compared to the overall frequency of the modal verb KÖNNEN, if compared to the
frequency counts for just the instances of KÖNNTE and KONNTE, we can see that this
confusion causes a substantial amount of errors when either of these two forms is
attempted. This error rate also remains stable across the three learner groups, so there
does not seem to be an improvement in the learners‟ ability to distinguish the two:
160
Year A Year B Year C
preterite KONNTE (raw figures) 18 53 16
Konjunktiv II KÖNNTE (raw figures) 24 41 48
KÖNNTE instead of KONNTE and vice versa 9 21 14
errors as % of all KONNTE + KÖNNTE 21.4 22.3 21.9
TABLE 31: ERROR RATE FOR NON-EPISTEMIC KÖNNTE/KONNTE
It has been pointed out before that out of all the modal verbs KÖNNEN is the one that
shows the greatest tendency for indeterminacy between epistemic and non-epistemic
meaning, especially in passive or impersonal “man/es” constructions or with certain main
verbs (e.g. mental and speech act verbs). This is why the annotation category „U‟
(unclear) is only really relevant in terms of frequencies for this modal verb. The
respective frequency counts for learner and native speaker groups are shown in Figure
10:
FIGURE 10: FREQUENCIES OF INDETERMINATE KÖNNEN (ANNOTATION CATEGORY “U”)
In these cases, different factors come together that pull in opposite directions in the
decision between epistemic and non-epistemic modality, or rather there are no clear
indicators present that favour one over the other reading. Passive and impersonal
(man/es) constructions avoid naming an explicit human agent, the verb sagen (say), for
example, is a performative verb but not a prototypical action verb (both factors that would
0.0
5.0
10.0
15.0
20.0
25.0
YEAR A YEAR B YEAR C KEDS
freq. per
10,000 words
161
disfavour a non-epistemic reading), but there are also no factors that would favour an
epistemic reading. Examples from the corpus are:
[c2031_05] Es gibt immer die Gefahr, dass Euthanasie kann missgeraucht werden. „There is always the danger that euthanasia can be misused.‟
[c1054_03] Andereseits kann man sagen, dass die Einführung von Benimm-Kursen an Schulen sinnvoll ist. „On the other hand one can say that the introduction of courses in good behaviour in schools is useful.‟
[a1018_03] Die Lernprogramme können auch Kinder mit Lernprobleme helfen. „The e-learning programs can also help children with learning difficulties.‟
[keds_059] Kreuzen sich diese Tiere mit wildlebenden Verwandten, so werden die manipulierten Gene weitergegeben, was fatale Folgen auf diese Spezies und damit auf das gesamte Ökosystem haben kann. „If these animals breed with their wild relatives, then the genetically engineered genes get passed on which can have fatal consequences for this species and therefore for the whole eco-system.‟
The reason why the frequency figures are much higher for Years A and C is that in
these two groups, the learners use those constructions more than the native speakers
that render the resulting utterances indeterminate (usually as mergers, see 2.3.1.1).
These are, in particular, combinations of man with mental and speech act verbs (see
example [c1054_03] above)112. This coincides with a general and proportionally constant
overuse of the verb sagen (say) in all learner groups (Year A: 10.6; Year B: 11.4; Year C:
10.3; KEDS: 4.0), so it seems that the high proportion of indeterminate uses of KÖNNEN
stems in part from the repetitive use of a limited range of vocabulary items in certain
formulaic patterns.
Apart from expressing possibility, KÖNNEN can also be used as an alternative to
DÜRFEN in the meaning of „having permission‟. As the relationship between these two is
revealing, I will deal with both of them together in the next section.
5.1.2.5 DÜRFEN and KÖNNEN (permission)
KÖNNEN has the English form/meaning equivalent CAN for expressing permission and
learners use it in this meaning:
112
And the verb helfen (help) (example [a1018_03] above), which is in parts topic related.
162
[b1003_06] Sie finden es besonders überraschenden, dass britische Studenten ihre Professoren mit dem Vornamen ansprechen können. „They find it particularly surprising that British students can address their professors by their first name.‟
[a2009_03] Das aktuellen Gesetz verlangt, dass Kneipen an 11.30 schliessen müssen, aber mit dem neuen Gesetz können sie für so lange wie sie wollen offen bleiben. „The current law demands that pubs close at 11.30 but with the new law they can stay open for as long as they want.‟
DÜRFEN has no English modal verb equivalent113. However, even Year A learners
have already integrated DÜRFEN into their modal system and actually use it more often
to express permission than KÖNNEN – a trend that is also visible in Year C and matches
the native speaker behaviour:
Year A Year B Year C KEDS
DÜRFEN (raw figures) 12 15 74 46
DÜRFEN (per 10,000 words) 2.6 2.3 8.4 5.6
KÖNNEN as permission (raw figures) 5 25 24 10
KÖNNEN as permission (per 10,000 words)
1.1 3.8 2.7 1.2
TABLE 32: DÜRFEN AND KÖNNEN AS PERMISSION
Another aspect where the learners show similar usage patterns to the native speakers
is negation, effectively turning permission into prohibition, as in the following example:
[c1047_02] Sachlich angesehen, dürfen Lehrerinnen theoretisch keinen persönlichen oder religiösen Einfluss auf die Schüler ausüben, denn ... ‟Objectively, teachers are not allowed to exert any kind of personal or religious influence on the pupils, because …„
For all other modal verbs negation occurs in only very small proportions of the overall
use, but Table 33 shows that in most of the subcorpora (including the native speakers‟)
DÜRFEN occurs more in negated contexts than in affirmative ones.
113
N.B. The modal verb MAY is available for expressing permission, but it is not a form/meaning equivalent of DÜRFEN. It is also mainly restricted to formal situations and furthermore to questions rather than affirmative statements. These restrictions do not apply to DÜRFEN.
163
Year A Year B Year C KEDS
DÜRFEN (raw figures) 12 15 74 46
negated DÜRFEN (raw figures) 6 6 46 29
negated DÜRFEN as % of all DÜRFEN 50.0 40.0 51.4 63.0
TABLE 33: NEGATED DÜRFEN
We have seen in section 5.1.2.2 that there is difficulty involved with negated
permission due to the L1 interference of MUST NOT, which leads to the erroneous use
of *NICHT MÜSSEN rather than NICHT DÜRFEN. These instances, in a pragmatic
sense, are clearly intended to denote prohibition (but expressed in the wrong linguistic
form). So, if we add them to the frequency for NICHT DÜRFEN, we can see that the
proportions of negative contexts for permissions (i.e. prohibitions) is actually very similar
for learners and native speakers and stable across the learner year groups:
Year A Year B Year C KEDS
DÜRFEN + *NICHT MÜSSEN (raw figures) 18 21 82 46
negated DÜRFEN + *NICHT MÜSSEN (raw figures) 12 12 46 29
negated DÜRFEN as % of all DÜRFEN 66.7 57.1 56.1 63.0
TABLE 34: NEGATED DÜRFEN INCLUDING *NICHT MÜSSEN
We can establish, then, that in a pragmatic sense the learners even in Year A have
incorporated the L2-specific DÜRFEN into their interlanguage modality system in a
native speaker like way. Learners in all years do have difficulties with the linguistic forms
for prohibition (i.e. NICHT DÜRFEN), but proportionally, they show the same tendency to
use DÜRFEN in negative contexts as the native speakers.
Yet another similarity between learners and native speakers is found when looking at
the collocating main verbs. The one verb that co-occurs more than any other with
DÜRFEN is vergessen (forget) in constructions like Wir dürfen nicht vergessen, dass…
or Man darf nicht vergessen, dass… (We/One must not forget that…) and this holds for
learners and native speakers alike (Year A: 7; Year B: 1; Year C: 11 instances), although
native speakers show more vocabulary variation than the learners as they also use
phrases like wir dürfen nicht außer Acht lassen or wir dürfen nicht vernachlässigen,
which express similar meanings as vergessen (KEDS: 8 instances).
164
[a2022_01] Auch , man darf nicht vergessen dass, der Einfluss nicht immer schlecht ist, zum Beispiel der Stars wie Princess Diana und von Comic Relief. ‟One must also not forget that this influence isn‟t always bad, for example that of stars like Princess Diana or of Comic Relief.‟
[c2024_04] Wir dürfen auch nicht vergessern, dass das Kaufen der Ostprodukte nicht die einzige Trend ist, die jetzt wieder populär wird. ‟We must also not forget that buying “Eastern” products is not the only trend that is becoming popular again now.‟
[keds_016] Doch bei all diesen Vorteilen und Chancen einer Fernreise in Dritte-Welt-Länder darf man nicht vergessen, dass es auch Probleme und Gefahren gibt. ‟But with all these advantages and opportunities of foreign travels into third-world-countries one must not forget that there are also problems and dangers.‟
Here, the learners have evidently acquired a typical phrase that native speakers use in
their argumentative writing in German. Also, a cross-check with the English native
speaker corpus LOCNESS reveals that the corresponding English phrases MUST NOT +
forget, although possible, is not used at all in the argumentative essays in that corpus114.
We are therefore not dealing with a case of L1 transfer, but rather with an „L2-internal‟
acquisition.
5.1.2.6 WOLLEN (volition)
WOLLEN is another L2-specific modal verb that is part of the learners‟ mastered
repertoire. It is consistently overused by a large margin in all learner groups compared to
the native speakers, but the reason for this is not, as we might expect, interferences with
the English modal verb WILL. Apart from only a handful of errors spread over the three
learner subcorpora, WOLLEN is always used correctly in its function of expressing the
subject‟s volition, intention or willingness.
[a2017_01] Wir wollen alle die Figur unserer Libelingsstar haben oder hübsch wie die berühmte Modelle sein. „We all want to have the figure of our favourite star or be beautiful like the famous models.‟
[keds_051] Sie wollen perfekt sein, allen Ansprüchen der Gesellschaft gerecht werden und durch ihre künstliche Schönheit in allen Lebensphasen brillieren.
114
A search in the British National Corpus revealed 53 instances of “must not forget”, which equates to a mere 0.54 instances per 1,000,000 words. The similar phrase “should not forget” returned 43 instances in the BNC and 1 in LOCNESS.
165
„They want to be perfect, meet all of society‟s expectations and shine in all phases of their lives through artificial beauty.‟
The only difference to the native speaker group that could be detected was that in
Year B there is a high number of 1st person singular subject sentences (ich) with
WOLLEN. This is largely due to one particular assignment (the intercultural project) in
this year group where students write about personal experiences (see Footnote 68 in
4.2.1). Other than that, the learners‟ use of WOLLEN does not differ qualitatively from
that of the native speakers – however, the quantitative differences are statistically highly
significant.
5.1.2.7 MÖCHTE (attenuated volition) and MÖGEN (affection/liking)
It is interesting to note that the overuse of WOLLEN does not coincide with an underuse
of MÖCHTE, which is the alternative (and generally considered a more tentative or
polite) form of expressing one‟s wish or intention. Rather the opposite is the case:
MÖCHTE is also overused in Years A and B; in Year C it is used with almost exactly the
same frequency as in KEDS. So, learners do not simply prefer to use WOLLEN in
volitional contexts whereas native speakers prefer MÖCHTE, but they tend to express
their own or other people‟s intentions or volition in their essays more often overall.
Looking at the data, it becomes obvious that MÖCHTE is also often used with a 1st
person singular subject and most of these instances are used for signposting purposes,
telling the reader how the text is structured:
[a2025_01] Abschließend möchte ich sagen, dass der Einfluss der Stars auf die Jugend ist zu groß geworden weil, ... „Finally, I would like to say that the influence of stars on the youth of today has become to big because …‟
[b1006_06] Zum Schluss kommen, möchte ich die Hauptpunkte wiederholen. „To come to a conclusion, I would like to reiterate the main points.‟
The following table shows the number of instances of 1st person singular subject and the
number and proportion of signposting utterances:
166
Year A Year B Year C KEDS
MÖCHTE (raw figures) 34 60 30 32
1st person (raw figures) 11 18 4 19
signposting (raw figures) 7 15 4 14
signposting as % of all MÖCHTE 20.6 25.0 13.3 43.8
TABLE 35: MÖCHTE IN 1ST
PERSON SIGNPOSTING UTTERANCES
The table reveals two interesting aspects. First of all, we can see that the use of
MÖCHTE as a text structuring device might explain partly why this modal verb is
overused in Years A and B but not in Year C, as in the latter group MÖCHTE is not used
much at all in this function. This in turn means, however, that with regard to signposting
Years A and B actually use MÖCHTE more like the native speakers, although in terms of
overall frequency of the modal verb alone, the Year C learner group seemed closer to
the native speakers. Even so, the native speakers actually use MÖCHTE in this function
roughly twice as much as the learners in Years A and B. On top of this, there is a
qualitative difference in the signposting phrases between the learners and the native
speakers insofar as the native speakers use a much broader range of different infinitives
and phrases with MÖCHTE than the learners. The latter repeatedly use different
combinations of the following phrase:
“Zum Schluss/Abschließend möchte ich sagen/betonen/behaupten…” (To conclude, I
would like to say/stress/claim…), whereas the former use a whole variety of
constructions, such as:
“Zum Schluss möchte ich auf … eingehen” (To conclude I would like to address …)
“Zusammenfassend möchte ich sagen …” (In summary I would like to say …)
“Im Folgenden möchte ich … erläutern” (In the following I would like to explain …)
“An dieser Stelle möchte ich einige Zahlen … aufzählen” (At this point I would like to cite
some figures …)
“Am Beispiel … möchte ich … begründen” (Using … as an example I would like to
account for …) etc.
The remainder of instances of MÖCHTE (with 3rd person subjects) in the learner data
does not differ qualitatively from that of the native speakers, but the number is
considerably higher.
With the indicative form MÖGEN, the overall frequency picture is similar to MÖCHTE
– overuse in Years A (16 instances; 3.5 per 10,000 words) and B (20; 3.0) and almost
167
equal use to native speakers in Year C (6; 0.7). As there are only two instances in KEDS
(0.2 per 10,000 words), it is not possible to determine qualitative differences in the use of
MÖGEN between learners and native speakers. All that can be done is describe the
contexts in which the learners use the modal verb and infer that the difference to native
speakers lies in the fact that they do not use MÖGEN in these functions in their own
writing. First of all, we can establish that all uses of MÖGEN in the learner data (and,
incidentally, in the native speaker data) express liking/affection or (negated) dislike, but
never the more literary meaning of indicating that the subject is free to implement a
proposition should they wish to do so:
[a2025_03] Die Jugendlichen und Studenten mögen die Idee, dass mann für vier und zwanzig Stunden Alkohol trinken kann. „Young people and students like the idea of being able to drink twenty-four hours.‟
[b1018_05] Sie trinkt Alkohol nicht gern, weil sie mag nicht den Geschmack. „She doesn‟t like drinking alcohol, because she doesn‟t like the taste.‟
As stated before, MÖGEN is not greatly overused in Year C, so we will turn our
attention to Years A and B. In Year B the explanation for the overuse is relatively simple
as it is, again, largely due to the particular assignment of the intercultural project. In this,
the learners are required to ask international students about their experiences studying
in a foreign country. It is therefore not surprising that there are many instances in these
texts (see example [b1018_05] above) where students report what these students have
said that they like or dislike about life at a British university (14 out of 23 instances =
61%). The other instances are similar to those in Year A; they describe in often relatively
short, simple sentences what the person or group of persons that is currently being
discussed likes or does not like:
[a1023_04] Jetzt mögen immer mehr Mädchen Computers auch. „Now, more and more girls also like computers.‟
[a2014_01] Die Jugend *magen (target: mag) die Stars sehr gern. „Young people like the stars very much.‟
[a1011_03] Es gibt eine Gruppe, die nicht die aktuelle Lage mögen. „There is a group that doesn‟t like the current situation.‟
This way of using MÖGEN captures the main function of this modal verb in the learner
data, but we can see that its frequency continuously declines from Year A to Year C,
where it sinks to an almost native speaker-like level. Again, we will see parallels to this
168
trend in the analysis of modal adverbials expressing the same meaning in the next
section (Section 5.2.3). The reason for this overuse can only be speculated about.
Figures from the LOCNESS corpus indicate that there might be L1 influence at work, as
similar expressions of liking or dislike are more prominent in the native English essay
writing: “like” occurs 41 times in this function, which makes 2.0 instances per 10,000
words as compared to only 0.2 instances of MÖGEN in this function in KEDS.
This concludes the analysis of non-epistemic modal verbs. In this category, we have
seen that the learners are generally able to use all different modal verb meanings,
regardless of whether they have English form/meaning equivalents or not. In this
respect, it could be argued that all of these meanings are part of the learners‟ standard
repertoire of modal means. At closer inspection, however, it has become obvious that
there are often finer differences in the types of constructions and contexts in which
learners and native speakers employ the respective modal verbs. On the one hand, this
shows that the learners‟ modal interlanguage system is often not as native speaker-like
as it might seem at first glance. One of the issues that play a significant role here is – as
we have already seen for epistemic modal verbs – the ability, or rather inability, to
distinguish and appropriately deal with subjunctive verb forms; other factors are
collocation and colligation patterns. On the other hand, it is often exactly in those points,
where the learner‟s interlanguage system differs from the native speakers‟, that we are
able to see a marked progression towards an approximation of the native speaker‟s
modal system from the lesser to the more advanced learner year groups.
5.2 PERIPHRASTIC MODAL EXPRESSIONS
5.2.1 Overview
Overall, 2180 periphrastic modal expressions (modal adverbials, quasi-modal verbs and
modal lexical verbs) were annotated in CLEG and 924 in KEDS. The distribution of
periphrastic modal expressions (PMEs) across the three learner year groups is shown in
Table 36, along with proportional frequency figures:
Periphrastic modal expressions Year A Year B Year C KEDS
raw figures of PMEs 426 691 1063 924
169
per 10,000 words 92.3*** 104.9 122.2 112.6
TABLE 36: OVERALL FREQUENCIES OF PERIPHRASTIC MODAL EXPRESSIONS IN LEARNER AND NATIVE
SPEAKER CORPORA (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
Only in Year A is the underuse of periphrastic modal expressions statistically highly
significant compared to the native speaker corpus KEDS. The slight underuse in Year B
and overuse in Year C compared to KEDS are not statistically significant. It has been
shown in the previous section that modal verbs are overused in all three learner groups
compared to the native speakers. This means that only in Year A is the significant
overuse of modal verbs counterbalanced by a significant underuse of periphrastic modal
expressions. The following chart displays the comparison between modal verbs and
PMEs more clearly:
FIGURE 11: COMPARISON OF MODAL VERBS AND PERIPHRASTIC MODAL EXPRESSIONS (PMES)
Figure 11 shows that there is a global tendency by the learners to overuse modality
markers, as the underuse of PMEs in Years A and B is much smaller than the overuse of
modal verbs (and in Year C PMEs are overused anyway). This global overuse of modal
expressions is statistically highly significant for Years A and C. The figures do not show a
significant overuse in Year B. As all the other variables are the same for this group as
the other two groups, this has to be attributed to the slightly different make-up of texts in
0.0
50.0
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
Year A Year B Year C KEDS
freq. per
10,000 words
PMEs
modal verbs
170
this year group (see Footnote 68). This is a further indication of just how text-type
sensitive the use of modal expressions is.
The overuse of modal expressions in argumentative texts also shows if we compare
epistemic and non-epistemic markers (modal verbs and PMEs together) across the year
groups and KEDS (Figure 12 below). We can clearly see that, overall, both epistemic
and non-epistemic modal expressions are generally overused by the learners115.
FIGURE 12: OVERALL FREQUENCIES OF EPISTEMIC AND NON-EPISTEMIC MODAL EXPRESSIONS
However, only a much more detailed analysis will reveal what the relationships really
are between patterns of modal verb use and periphrastic modal expressions.
We have seen in the theoretical outline (2.3.2 and 2.3.3) that, unlike the small set of
modal verbs, the group of periphrastic modal expressions comprises a large open set of
items of varying word class and syntactic makeup. Whereas for the modal verbs it was
important to distinguish which of their many different functions the learners used, there
are different aspects that are relevant for the periphrastic modal expressions. First of all,
as the set of investigated items is so large, a point to explore is how many of these items
115
Modal infinitives and Konjunktiv II are not included in these figures as they will only be discussed later on. Their inclusion would, however, not affect the overall observation.
0.0
50.0
100.0
150.0
200.0
250.0
Year A Year B Year C KEDS
freq. per
10,000 words
epistemic modality
non-epistemic modality
171
are actually used by the learners and native speakers. As laid out in the methodology
section (4.3.1), the list of annotated periphrastic modal expressions contains 227
items116. Table 37 provides an overview of the lexical diversity for PMEs by giving the
number of different types that are used in each of the subcorpora, together with a type-
token ratio117 for this group of expressions:
Year A Year B Year C KEDS
number of PME types 58 96 106 142
raw figures of PMEs (freq. per 10,000 words in brackets)
426 (92.3) 691 (104.9) 1063 (121.3) 924 (112.6)
type/token ratio for PMEs 13.6 13.9 10.0 15.4
TABLE 37: LEXICAL DIVERSITY OF PERIPHRASTIC MODAL EXPRESSIONS
We can see that the steady increase in the proportional frequencies of PME use is
matched by a steady upward progression in the number of different PME types that
occur in the learner subcorpora from Year A to Year C. This means, unsurprisingly yet
reassuringly, that the learners during their study acquire a broader range of vocabulary
to choose from in order to express modal meanings. Even in Year C, however, the
number of PME types only reaches about three-quarters of that of the native speaker
group, despite an overuse of PMEs in Year C compared to KEDS118. Curiously, the
type/token ratio for Years A and B is not too dissimilar to the native speakers‟, whereas it
is considerably lower for Year C. This means that in Years A and B, students use a more
limited range of periphrastic modal expressions while at the same time not using them as
often as native speakers, too. In Year C, the learners also make use of a smaller range
of items than the native speakers, but they use this smaller set more often than the
native speakers, i.e. there is more repetition of PMEs in Year C.
The same distinction between epistemic and non-epistemic expressions has been
made as with the modal verbs. Within this dichotomy, the periphrastic modal expressions
have been assigned to the same semantic categories as the modal verbs, with the
additional categories of assertive and negative modal expressions (see 2.3.2.2 and
116
Productive morphemes like *bar and *weise count as one item, although they, of course, produce several different types.
117 N.B. for technical reasons it was not possible to calculate standardised type/token ratios that account for differences in text lengths in the three sub-corpora.
118 In the general native speaker corpus LIMAS 356 different PME types were counted, indicating that the range of productively used PMEs is still relatively low for the novice writers in KEDS as compared to professional writers.
172
4.3.2). Figure 13 displays how epistemic and non-epistemic PMEs are distributed over
the four subcorpora:
FIGURE 13: EPISTEMIC VS. NON-EPISTEMIC PMES
From this figure we can determine that, in terms of proportions of distribution, the
learners and native speakers display similar behaviour: epistemic PMEs make up the far
bigger share of all PMEs, non-epistemic expressions are used less. (This counteracts
exactly the behaviour with modal verbs, where non-epistemic modal verbs are more
frequent than epistemic ones and it corroborates the statement that, in German, PMEs
play a more prevalent role in epistemic modality than modal verbs.) In Years A and B,
the epistemic/non-epistemic ratio is higher compared to that of Year C and the native
speaker group: Year A: 84% epistemic; Year B: 80%; Year C: 73%; KEDS: 72%. This
comes about because of an underuse of non-epistemic PMEs in these two year groups
compared to the native speakers. In Year C, the epistemic/non-epistemic PME ratio is
almost identical to that of KEDS as both epistemic and non-epistemic PMEs are slightly
overused. At this level, the behaviour of learners and native speakers appears to be
relatively similar. A more in-depth analysis, however, will reveal significant differences.
How these differences are manifest in the various semantic categories across the three
learner groups will be discussed in the following sections.
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
Year A Year B Year C KEDS
freq. per
10,000 words
epistemic PMEs
non-epistemic PMEs
173
5.2.2 Epistemic periphrastic modal expressions
5.2.2.1 Assertive periphrastic modal expressions
Assertive expressions are the first group of PMEs that will be explored in detail. As
discussed in section 2.3.2.2, these are expressions that either emphasise the assertion
of a statement, add some evaluative dimension to the assertion or focus the assertion on
inter-subjective evidence. There are six categories in this group:
Purely assertive, evaluative assertive (EA)-positive, EA-reservedly positive, EA-
negative, EA-other dimensions, evidence-focused assertive. Figure 14 provides an
overview of their distributions across the learner and native speaker corpora:
FIGURE 14: FREQUENCIES OF ASSERTIVE PMES
We can see that purely in terms of frequencies the respective learner groups employ
assertive PMEs, with a few exceptions, in a similar fashion to the native speakers. A
closer look at the individual items will reveal if this is also the case for the other relevant
aspects that have been identified.
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
18.0
freq. per
10,000 words
Year A
Year B
Year C
KEDS
purely assertiveEA-positiveEA-reservedly
positive
EA-negativeEA-other
dimensions
evidence-
focused
assertive
174
The following purely assertive PMEs occur in the subcorpora119:
Year A Year B Year C KEDS
wirklich 11
(2.4) wirklich
15 (2.3)
wirklich 82
(9.4) wirklich
35 (4.3)
tatsächlich 2
(0.4) tatsächlich
2 (0.3)
tatsächlich 22
(2.5) tatsächlich
6 (0.7)
gewiss 1
(0.1)
TOTAL 13
(2.8) TOTAL
17** (2.6)
TOTAL 105*** (12.0)
TOTAL 41
(5.0)
TABLE 38: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PURELY ASSERTIVE PMES (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
While, proportionally, the frequencies for Years A and B are only slightly below that of
KEDS, there is a highly significant overuse of purely assertive PMEs in Year C.
Wirklich is the adverbial that is used most often by both learners and native speakers. All
learner groups use wirklich mostly in appropriate contexts. The only qualitative difference
that could be discovered was that in KEDS, ca. 30% of instances of wirklich occur in
direct or indirect questions, a tendency that is also present to the same extent in Year C,
but less in Years A (20%) and B (13%):
[c2031_01] Wissen Sie wirklich, wie Ihr Frühstück auf Ihren Teller kommt? „Do you really know where your breakfast comes from?‟
[c1049_04] Obwohl diese Filme gewalttätige Szene enthalten, ist es fraglich, ob sie wirklich für die Tragödien verantwortlich sind. „Although these films contain scenes of violence it is doubtful that they really are to blame for the tragedies.‟
As for tatsächlich, the concordances reveal that all four instances in Years A and B
are actually not quite used appropriately. The learners used them in the meaning of in
fact which is used in „corrective‟ contexts, i.e. to say that the common believe is X, while
– in fact – the truth is Y, e.g.:
[a1005_02] obwohl Autobahnen berühmt für keine Geschwindigkeitsbegrenzung sind, obwohl tatsächlich es gibt jetzt begrenzungen an viele Stellen. „although motorways are famous for no speed limit, although in fact there are now limits in many places.‟
119
For a list of all items that were annotated in the different PME categories see Appendix A.
175
In German, this is usually expressed by eigentlich or in Wirklichkeit, as the speaker
relativises the validity of the preceding proposition (cf. Dalmas 1999). In Year C,
however, the learners use tatsächlich in the same way as the native speakers.
[c2043_01] Ein Punkt, der ganz berechtigt ist, ist ob solche frühe Diagnose von Krankheiten tatsächlich hilfreich sind. „One valid point is whether this early diagnosis of diseases is actually helpful.‟
[keds_059] Und die Gentechnik bietet ein unerschöpfliches Arsenal, biologische Waffen noch effektiver zu machen [...] was nach billigstem Science Fiction klingt , ist tatsächlich schon Realität. „And genetic engineering offers an unlimited arsenal of means to make biological weapons even more effective […] what sounds like trashiest science fiction is actually already reality.‟
Positive evaluative-assertive PMEs (EA-positive) only occur in very small numbers
which are distributed across the subcorpora as follows:
Year A Year B Year C KEDS
zum Glück 2
(0.4) glücklicherweise
4 (0.6)
glücklicherweise 3
(0.3) glücklicherweise
1 (0.1)
glücklicherweise 1
(0.2)
*dankbar (target: Gott sei dank)
1 (0.2)
zum Glück 1
(0.1)
Gott sei dank 1
(0.1)
TOTAL 3
(0.7) TOTAL
5 (0.8)
TOTAL 5
(0.6) TOTAL
1 (0.1)
TABLE 39: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF POSITIVE EVALUATIVE-ASSERTIVE PMES
The frequency figures in this group are too low for significance testing but they indicate
that native speakers seem to avoid positive evaluative-assertive PMEs in argumentative
writing while learners are not so reluctant to incorporate them. A reason for this might be
that these PMEs signal subjective speaker involvement, which might be considered
unhelpful in the supposedly rational (objective) reasoning in argumentative writing.
However, in the next category we can see that native speakers do use evaluative
PMEs, but they prefer the more reservedly positive ones:
176
Year A Year B Year C KEDS
zumindest 2
(0.4) zumindest
4 (0.6)
immerhin 4
(0.5) zumindest
22 (2.7)
wenigstens 1
(0.2) wenigstens
1 (0.1)
wenigstens 5
(0.6)
immerhin 4
(0.5)
TOTAL 3*** (0.7)
TOTAL 4*** (0.6)
TOTAL 5*** (0.6)
TOTAL 31
(3.8)
TABLE 40: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF RESERVEDLY POSITIVE
EVALUATIVE-ASSERTIVE PMES (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
Curiously, these PMEs are underused by the all learner groups, which is manifest in both
frequencies (statistically significant for all groups) and lexical richness. On top of this,
there are qualitative differences. From the concordances it can be seen that the native
speakers use zumindest and wenigstens to add a reservedly positive evaluation in
constrictive, qualificatory contexts:
[keds_018] Insofern können Frauen auch selbst entscheiden, ob sie die Chance der Gleichberechtigung beim Bund annehmen oder nicht. Aber sie haben nun wenigstens ein Recht darauf. „In this sense the women can decide for themselves whether they take up the chance of equal opportunities in the Armed Forces or not. But at least they now have the right to do so.‟
[keds_058] Auch Erbkrankheiten könnten mit Hilfe der Gentechnik geheilt, oder zumindest die Veranlagung sehr früh erkannt werden. „Hereditary diseases could also be cured with the help of genetic engineering, or at least the predisposition could be detected very early on.‟
In Year A, all three instances of wenigstens and zumindest are used incorrectly in the
quantifying sense of at least (target: mindestens), in Year B three out of four are used in
this sense:
[b1011_06] Jede hat viele Freunden, die sich mit ihnen* zumindest (target: mindestens) dreimal pro Woche treffen. „All of them have many friends who meet with them at least three times a week.‟
This leaves only one correct instance of wenigstens in Year C and one of zumindest in
Year B:
177
[b1025_05] Jedoch, fast alle Deutschen trennen zumindest Papier und Glas, entweder in speziellen Mülleimern oder ... „But almost all German at least separate paper and glass, either in special bins or …‟
The category of negative evaluative-assertive PMEs is almost exclusively made up of
instances of leider (unfortunately) in the corpus data and the relative frequencies are
virtually equal in all learner and native speaker subcorpora (no statistically significant
differences):
Year A Year B Year C KEDS
leider 13
(2.8) leider
15 (2.3)
leider 23
(2.6) leider
21 (2.6)
traurigerweise 1
(0.2)
TOTAL 13
(2.8) TOTAL
16 (2.4)
TOTAL 23
(2.6) TOTAL
21 (2.6)
TABLE 41: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF NEGATIVE EVALUATIVE-ASSERTIVE PMES
The learners use leider in the same way as the native speakers, including a certain
propensity to use it sentence-initial, thus giving it a thematised, focused position. This is
proportionally much higher for Years A and B but drops to the same level as the native
speakers in Year C:
Year A Year B Year C KEDS
sentence-initial leider (raw figures) 9 8 9 8
as % of all instances of leider 69.2 53.3 39.1 38.1
TABLE 42: SENTENCE-INITIAL LEIDER
The last category in the group of evaluative-assertive PMEs consists of idiosyncratic
items (typically only occurring once each) that express evaluation with reference to a
particular dimension (other than positive/negative). The following items were found in the
corpora:
178
Year A Year B Year C KEDS
verständlicherweise üblicherweise ökonomischerweise brisanterweise
paradoxerweise freundlicherweise berechtigterweise paradoxerweise
*falschlicherweise typischerweise interessanterweise unvernünftigerweise
unnötigerweise
unvermeidlicherweise
TABLE 43: EVALUATIVE-ASSERTIVE PMES ALONG OTHER DIMENSIONS
It is interesting to note that, in this category, Year C actually surpasses the native
speakers in terms of lexical diversity, demonstrating the learners‟ understanding of the
morphological process for forming these adverbs and their lexical creativity.
Finally, the last category in this first group of assertive PMEs comprises evidence-
focused PMEs. This means that these PMEs aim to imply a subject-external assertion of
a given utterance that is based on some kind of objective evidence (although the exact
nature of this evidence remains, in most cases, unspecified). It is therefore not surprising
that, in the argumentative texts that are being investigated here, this is the biggest
category within the group of assertive PMEs, both in terms of frequencies and lexical
diversity (the frequency differences for learners and native speakers are all not
statistically significant):
179
Year A Year B Year C KEDS
natürlich 27
(5.9) natürlich
35 (5.3)
Tatsache 35
(4.0) natürlich
45 (5.5)
Tatsache 16
(3.5) Tatsache
17 (2.6)
natürlich 31
(3.5) Tatsache
28 (3.4)
nicht leugnen
5 (1.1)
nicht leugnen 10
(1.5) nicht leugnen
27 (3.1)
offensichtlich 9
(1.1)
es steht fest
4 (0.9)
offensichtlich 9 (1.4)
ohne/kein/ außer Zweifel
16 (1.8)
ohnehin 9 (1.1)
ohne/kein Zweifel
3 (0.7)
ohne/kein/ außer Zweifel
8 (1.2)
es steht fest 10
(1.1) sowieso
7 (0.9)
offensicht-lich
1 (0.2)
offenkundig 6
(0.9) offensichtlich
10 (1.1)
bekanntlich 6
(0.7)
bewiesen 1
(0.2) zweifellos
4 (0.6)
selbstver-ständlich
5 (0.6)
zweifellos 4
(0.5)
bewiesen
2 (0.3)
sowieso 4
(0.5) bewiesen
3 (0.4)
in jedem Fall 2
(0.3) bekanntlich
2 (0.2)
selbstverständ-lich
3 (0.4)
sowieso 2
(0.3) zweifellos
2 (0.2)
zwangsläufig 2
(0.2)
bekanntlich 1
(0.2) ohnehin
1 (0.1)
erwiesener-maßen
2 (0.2)
selbstver- ständlich
1 (0.2)
unbestreit- bar
1 (0.1)
zweifelsohne 2
(0.2)
es steht fest 1
(0.2) unstreitbar
1 (0.1)
nicht leugnen 2
(0.2)
zwangsläufig 1
(0.2) unwiderleg-bar
1 (0.1)
es steht fest 2
(0.2)
zweifelsohne 1
(0.2) ohne Zweifel
1 (0.1)
eh 1
(0.1)
bekannter- maßen
1 (0.1)
in jedem Fall 1
(0.1)
auf alle Fälle 1
(0.1)
nachweislich 1
(0.1)
unbestreitbar 1
(0.1)
TOTAL 57
(12.4) TOTAL
100 (15.2)
TOTAL 146
(16.7) TOTAL
131 (16.0)
180
TABLE 44: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF EVIDENCE-FOCUSED
ASSERTIVE PMES
In this category, the native speakers clearly demonstrate the broadest lexical range,
although Year B learners already show a much broader range than Year A learners. The
top five items in each subcorpus make up the majority of occurrences in all subcorpora
(Year A: 96%; Year B: 80%; Year C: 81%; KEDS: 74%). Comparing these five items in
each subcorpus we can see that the first two are the same for all learner groups and the
native speakers, namely natürlich (of course) and expressions involving the noun
Tatsache (fact).
The concordances reveal that the adverb natürlich is used by all learner groups in the
same way as by the native speakers. It is already a firmly acquired item in Year A to
express the notion of emphasising the validity of an utterance by portraying it as a known
fact that will earn agreement from the reader („of course‟). Due to its function as a
sentence adverb, there is a great deal of freedom with respect to the position of natürlich
in the sentence and the learners make use of this as much as the native speakers,
varying between sentence-initial and sentence-internal positions:
[a2012_04] Natürlich wollten die Mehrzahl der Ausländer nicht zurück gehen, weil jetzt hatten sie Familie und Freunde in Deutschland. „Of course the majority of foreigners did not want to go back, because now they had their family and friends in Germany.‟
[b1018_04] Unter den verschiedenen Leuten, die bei globalen Firmen arbeiten, wird es heutzutage natürlich akzeptiert, dass arbeiten wichtiger als leben ist. „Amongst the different people that work with global companies it is nowadays of course accepted that work is more important than free time.‟
[c2029_02] Dieses System ist natürlich besser für Steuerzahler, weil sie nicht für die Studenten verantwortlich fühlen. „This system is of course better for the tax payer, because they do not feel responsible for the students.‟
For constructions with the noun Tatsache, the patterns that can be observed are more
different for learners and native speakers. The colligational sequence that occurs most
often is Tatsache directly followed by a subordinate dass-clause (that-clause), specifying
exactly what this „fact‟ is, for example:
[b1006_05] Die Tatsache, dass jeder Arbeiter sich auf das Ende des Tages freut, ist weiteren Beweis für die negative Einstellung zur Arbeit. „The fact that every worker looks forward to the end of the day is another piece of evidence for the negative attitude towards work.‟
181
This is the case for both learners and native speakers, although it is more pronounced
for the learners in Years A and C:
Year A Year B Year C KEDS
sequence: …Tatsache, dass… 13 10 32 18
as % of all instances of Tatsache 81.3 58.8 91.4 64.3
TABLE 45: COLLIGATIONAL SEQUENCE …TATSACHE, DASS…
However, a look at the further extension of this sequence reveals that in Year C, it has
developed into an almost fixed phrase: Es ist eine Tatsache, dass… (It is a fact that…):
[c1049_04] Es ist eine Tatsache, dass die Graphiken in Computerspielen immer mehr wie die Realität aussehen... „It is a fact that the graphics in computer games look more and more like reality …‟
[c1054_02] Es ist weiter eine Tatsache, dass das Kopftuch manchmal mehr als nur ein religiöses Symbol sein könnte. „It is also a fact, that the head scarf could sometimes be more than just a religious symbol.‟
This type of phrase occurs 14 times in Year C (44% of instances of the sequence
…Tatsache, dass…), but not at all in KEDS – and, curiously, also not in Years A and B.
In KEDS, the sequence …Tatsache, dass… is in most instances (61%) syntactically
integrated into the sentence as a prepositional phrase, e.g.
“…auf Grund der Tatsache, dass…” (…due to the fact that…)
“Die Begründung ... liegt in der Tatsache, dass...“ (The reason ... lies in the fact that...)
“Er argumentiert mit der Tatsache, dass…” (He argues with the fact that…)
“Die Diskrepanz … wird durch die Tatsache, dass… noch verstärkt.” (The discrepancy is
further acerbated by the fact that…)
In the other instances, the sequence typically occupies the subject position of a main
clause that often entails further embedded clauses, e.g.
“Die Tatsache, dass es Widerstand gab, wirft die Frage auf: …” (The fact that resistance
did exist raises the question: …)
“Aber die Tatsache, dass…, läßt hoffen, dass...” (But the fact that … gives rise to the
hope that…)120
120
This type of construction also occurs in Year C, but only in ca. 25% of …Tatsache, dass… occurrences as opposed to 44% in KEDS.
182
This means that the presented „fact‟ is not the only information conveyed in the
utterance, but rather is supplied as additional information on top of the surrounding
content of the sentence, which is conveyed through the predicate and other
complements of the main clause or subordinate clauses. The Year C sequence Es ist
eine Tatsache, dass… constitutes a matrix clause in itself (with the semantically empty
dummy-subject + copula verb construction es ist…), which does not allow for further
information to be conveyed in the same clause. This means that, in this syntactically
extrapolated construction, the assertion that the presented proposition is a fact is much
more thematised or foregrounded than in the more syntactically integrated constructions
of the native speakers.
Interestingly, learners in Years A and B predominantly employ assertions with the
noun Tatsache in the less formulaic way similar to the native speakers, i.e. syntactically
integrated and not thematised as either a prepositional phrase or the subject of the
sentence, e.g.:
[a1010_04] Ein Erziehungs Experte sagt, dass deutsche Schule Computers nicht so viel benutzen, trotz der Tatsache, dass Experte auch einig sind, dass Computers Kinder mit Lernproblemen helfen würden. „One education expert says that German schools do not use computers that much despite the fact that experts also agree that computers would help children with learning difficulties.‟
[b1038_01] Die Tatsache, dass Leute illegale Zeitungen lasen, beweist, dass die Pressefreiheit nicht nur für die Journalisten wichtig ist, aber auch für die Leserschaft der Presse. „The fact that people read illegal papers proves that freedom of press is not just important to journalists but also to the readers of the press.‟
The tendency by learners in Year C to repeatedly use formulaic sequences is also
evident when looking at the other top five items in this category. For ease of reference,
the top section of Table 44 is repeated here:
183
Year A Year B Year C KEDS
natürlich 27
(5.9) natürlich
35 (5.3)
Tatsache 35
(4.0) natürlich
45 (5.5)
Tatsache 16
(3.5) Tatsache
17 (2.6)
natürlich 31
(3.5) Tatsache
28 (3.4)
nicht leugnen 5
(1.1) nicht leugnen
10 (1.5)
nicht leugnen 27
(3.1) offensichtlich
9 (1.1)
es steht fest 4
(0.9) offensichtlich
9 (1.4)
ohne/kein/außer Zweifel
16 (1.8)
ohnehin 9
(1.1)
ohne/kein Zweifel
3 (0.7)
ohne/kein/außer Zweifel
8 (1.2)
es steht fest 10
(1.1) sowieso
7 (0.9)
TABLE 46: TOP FIVE EVIDENCE-FOCUSED ASSERTIVE PMES
We can see that, with the exception of offensichtlich (obviously) in Year B, after natürlich
and Tatsache the same items appear in all three learner groups in only slightly different
order and these items are all different from the ones that are ranked third to fifth highest
in the native speaker corpus. If we look more closely at the individual occurrences, it
becomes obvious that the three items nicht leugnen (cannot deny), es steht fest (it is
certain that) and ohne/kein/außer Zweifel (without a doubt) all lend themseves to the kind
of formulaic sequences that we have seen being used in Year C with the noun Tatsache,
and it is again in this advanced year group that the patterns seem to become fossilised.
The verb nicht leugnen occurs in all learner groups in the following sequences (all
meaning roughly it cannot be denied that…):
“Es ist nicht zu leugnen, dass…”
“Es lässt sich nicht leugnen, dass...“
“Man kann nicht leugnen, dass…”
“Niemand kann leugnen, dass…”
In Years A and B the frequency of this item is so low that the different sequences only
occur a few times each, thus giving the impression of a good variation of expressions. In
Year C, however, the strikingly predominant sequence is es ist nicht zu leugnen, which
actually makes up 21 of the 27 instances of nicht leugnen (78%). This marks a
considerable overuse compared to the mere two instances of nicht leugen in KEDS.
[c1057_03] Es ist nicht zu leugnen, dass Studiengebühren gegen arme Familie diskriminieren, ... ‟It cannot be denied that tuition fees discriminate against poor families…‟
A similar picture presents itself with ohne/kein/außer Zweifel. The sequences that are
used are:
184
“Es steht außer Zweifel, dass...“
“Es gibt keinen Zweifel, dass…”
“Ohne Zweifel…”
Again, the frequencies of these items are quite low overall for Years A and B with all
variations being used in roughly equal proportions. In Year C, however, the relative
overuse of expressions with Zweifel is juxtaposed by the dominance of just one phrase:
ohne Zweifel… (12 out of 14 instances, 86%). This expression is only used once in
KEDS.
[c2036_03] Ohne Zweifel leben wir in einer Welt, in der man nicht weit schauen muss, um eine Werbung zu sehen. ‟Without a doubt do we live in a world where you don‟t have to look far to see an advertisement.‟
Finally, es steht fest is also a fixed expression that the learners seem to have picked
up readily, especially in Year C, while not being used much by the native speakers (two
instances in KEDS). In addition, all instances of es steht fest in the learner data are
sentence-initial and used in exactly the same way, while in the native speaker data we
encounter two variations of this phrase:
“Somit steht fest, dass…”
“Fest steht jedenfalls, dass…”
[c1056_05] Es steht fest, dass Deutschland eine hochqualifizierte Arbeitskraft bietet. ‟It is certain that German offers a highly qualified work force.‟
In summary, we can say that in Years A and B the repertoire of items is relatively
small compared to Year C and KEDS. On top of this, the learners in these year groups
also use these items infrequently compared to the advanced learners and native
speakers; they do, however, use a broad variation of expressions and seem to be
syntactically more creative with them. In Year C, although a larger number of different
types can be found, the learners considerably overuse a small selection of items in a
very formulaic and fossilised way that does not reflect the native speakers‟ use. We have
also determined a propensity on the side of the learners, but most predominantly in Year
C, for positioning all of these items at the beginning of the sentence, thus giving the
evidence-focused assertive PMEs a prominent place within the information structure of
the utterance. The fact that this tendency for sentence-initial formulaic sequences only
185
becomes so overwhelming in Year C leads us to believe that we are not dealing with an
L1-transfer problem, but rather an L2-internal development in the learners‟ writing style
that diverges quite clearly from that of the native speakers. It will also be shown that
similar tendencies exist for PMEs in other categories.
5.2.2.2 Modally qualifying periphrastic expressions
Modally qualifying PMEs are those that have been matched up with the different
categories of modal verb meanings (see 2.3.2.2 and 4.3.2).
The following chart provides an overview of the distribution of these items across the
learner and native speaker groups:
FIGURE 15: FREQUENCIES OF MODALLY QUALIFYING PERIPHRASTIC EXPRESSIONS
As always, the overview chart gives a first impression of over- and underuse patterns.
From Figure 15 we can see that the PMEs in the most assertive category „compelling
conclusion‟ (the same category that the modal verb MÜSSEN belongs to) are
underused, while PMEs in the categories „subjective assumption‟ (modal verb DÜRFTE)
and „possibility‟ (modal verb KÖNNEN), which denote a lower degree of certainty, tend to
be overused. A more qualitative look into the individual categories will reveal any learner
specific patterns.
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
freq. per
10,000 words
Year A
Year B
Year C
KEDS
compelling
conclusion
tentative conclusionsubjective
assumption
possibilityevidentials
186
The following PMEs that have been associated with the category „compelling
conclusion‟ are found in the subcorpora:
Year A Year B Year C KEDS
überzeugt 8
(1.7) überzeugt
6 (0.9)
bestimmt 16
(1.8) sicherlich
28 (3.4)
bestimmt 3
(0.7) sicher
4 (0.6)
überzeugt 11
(1.3) sicher
21 (2.6)
sicherlich 2
(0.4) sicherlich
2 (0.3)
sicher 9
(1.0) mit Sicherheit
10 (1.2)
sicherlich 6
(1.8) überzeugt/ Überzeugung
8 (1.0)
mit Sicherheit 2
(1.1) bestimmt
5 (0.6)
TOTAL 12*** (2.8)
TOTAL 12*** (1.8)
TOTAL 44*** (5.0)
TOTAL 72
(8.8)
TABLE 47: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING A
COMPELLING CONCLUSION (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
Overall, the underuse of this category of items is statistically highly significant for all
learner groups compared to KEDS. Within the category, there are also qualitative
differences. The expression ich bin überzeugt (I am convinced) figures prominently in all
learner groups compared to a relatively small number in KEDS. There are, in fact, only
three occurrences of this phrase in KEDS, the rest are variations with the noun
Überzeugung, e.g. Ich bin der Überzeugung (I am of the conviction). In all learner groups
the basic sequence is often expanded into the following:
“Ich bin davon überzeugt, dass...” (I am convinced of it that…),
which is occasionally further modified with fest or persönlich:
“Ich bin fest davon überzeugt, dass...” (I am absolutely convinced that…)
“Ich bin persönlich davon überzeugt, dass...” (I personally am convinced that...)
Both of these additional modifications also occur in KEDS. The proportion of this
sequence in sentence-initial position is ca. 50% for each learner group and 33% for
KEDS, showing a similar pattern to the formulaic phrases discovered in the previous
section.
Furthermore, it is interesting to note that the learners in Years A and B clearly prefer
the clausal expression that signals the source of the epistemic necessity as speaker-
internal (first-person subject) over the adverbials sicherlich/sicher/mit
Sicherheit/bestimmt (certainly, surely), which are unmarked with respect to speaker-
187
involvement. Although more of these PMEs are used in Year C, the proportion of
statements of personal conviction is still much higher than in the native speaker group
(Year C: 33% vs. KEDS: 12.5%).
A similar observation can be made when looking at the next two categories –
„subjective assumption‟ and „tentative conclusion‟, respectively. As explained in section
4.3.2, the items in these two categories carry a similar degree of certainty, but differ in
terms of the source of epistemic authority. It is characteristic of the items in the category
„subjective assumption‟ that they explicitly identify the source of the epistemic authority
as subject-internal (as the modal verb DÜRFTE also implies the direction of the subject),
e.g. meiner Meinung nach (in my opinion). The ones in the category „tentative
conclusion‟ are neutral with respect to the authoritative source, e.g. wahrscheinlich
(probably).
Interestingly, in a direct comparison of the frequencies for these two categories (see
Figure 16), we can see that learners in Years A and B distinctly prefer the items that
explicitly express speaker-involvement („subjective assumption‟) over the neutral ones
(„tentative conclusion‟), whereas the two categories are more balanced in Year C and
KEDS (where items denoting impersonal tentative conclusions are actually preferred).
188
FIGURE 16: FREQUENCIES OF PME CATEGORIES „SUBJECTIVE ASSUMPTION‟ VERSUS „TENTATIVE
CONCLUSION‟
Moreover, there is not only an overuse of items in the category „subjective
assumption‟ compared to the category „tentative conclusion‟ in Years A and B, but also a
statistically highly significant overuse of these items compared to the native speaker
group. We will look into this category in more detail first. The following items occur in the
data:
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
Year A Year B Year C KEDS
freq. per 10,000 words
subjective assumption
tentative conclusion
189
Year A Year B Year C KEDS
meine Meinung
121
68 (14.7)
ich denke 61
(9.3) meine Meinung
46 (5.3)
meine Meinung 31
(3.8)
ich glaube 52
(11.3) ich glaube
55 (8.4)
ich glaube 27
(3.1) ich denke
14 (1.7)
ich denke 36
(7.8) meine Meinung
42 (6.4)
ich denke 17
(1.9) ich glaube
10 (1.2)
meine Ansicht
1 (0.2)
meine Ansicht 4
(0.6) meine Ansicht
1 (0.1)
meine Ansicht 6
(0.7)
meine Auffassung
1 (0.2)
meine Erfahrung
1 (0.2)
meine Auffassung
5 (0.6)
TOTAL 158*** (34.2)
TOTAL 162*** (24.6)
TOTAL 91
(10.4) TOTAL
66 (8.0)
TABLE 48: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING
SUBJECTIVE ASSUMPTIONS (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
Table 48 shows that there is virtually no difference in terms of lexical diversity between
the native speakers and all of the learner groups (although in KEDS the individual
occurrences are spread slightly more evenly across the different items). Also in terms of
the order of items there is great similarity between all subcorpora. We shall look at the
most frequent items in more detail. These are ich denke (I think), ich glaube (I believe)
and meiner Meinung nach/ich bin der Meinung (in my opinion).
We have already seen in the group of assertive PMEs that the learners show a
distinct tendency for using a limited set of expressions in a formulaic way. This can also
be observed with the items under scrutiny here. The expressions ich denke, ich glaube
and ich bin der Meinung can all be followed by a subordinate dass-clause and this is how
it is mostly used by the learners, but not by the native speakers, who prefer the
connection without dass:
[b1002_05] Deshalb denke ich dass die Institution von Ehe wenige wichtig für die Gesellschaft ist aber für viele Leute ist die Idee eine Ehe auf Lebenszeit sehr wichtig... ‟Therefore I think that marriage as an institution is less important for society as a whole but for may people the idea of marriage for life is very important…‟
[keds005] Ich denke, an den aufgeführten Argumenten kann man die positive Seite dieses Projekts sehen.
121
This encompasses all expressions that include the noun Meinung (opinion) and either a first person subject or a first person possessive pronoun, i.e. „ich bin der Meinung‟, „meiner Meinung nach‟, „ich vertrete die Meinung‟, „meine Meinung ist‟ etc. The same goes for expressions with Ansicht, Auffassung (opinion) and Erfahrung (experience).
190
‟I think from the arguments listed above the positive side of this project are visible.‟
Year A Year B Year C KEDS
sequence: ich denke, dass… 28 56 12 5
as % of all instances of ich denke 77.8 91.8 70.6 35.7
sequence: ich glaube, dass… 40 44 21 2
as % of all instances of ich glaube 76.9 80.0 77.8 20.0
sequence: ich bin der Meinung, dass…
29 10 14 7
as % of all instances of ich bin der Meinung
100.0 76.9 82.3 100.0
total sequences: ...denke/glaube/Meinung, dass
97 110 47 14
total as % of all instances 82.9 85.3 77.0 45.1
TABLE 49: COLLIGATIONAL SEQUENCES ICH DENKE, DASS…/ICH GLAUBE, DASS.../ICH BIN DER
MEINUNG, DASS…
On top of this, the learners exhibit a distinct tendency to position these expressions
prominently at the beginning of the sentence (ca. two-thirds of all instances), which,
again, is not shared to that extent by the native speakers:
Year A Year B Year C KEDS
sentence-initial ich glaube/ich denke/meine Meinung
104 103 55 24
as % of all instances of ich glaube/ich denke/meine Meinung
66.7 65.6 61.1 43.6
TABLE 50: SENTENCE INITIAL ICH DENKE/ICH GLAUBE/MEINE MEINUNG
Drawing these pieces of information together, it can be established that the learners
show a preference for expressing assumptions that explicitly identify the speaker as the
epistemic source122by using sentence-initial formulaic phrases followed by dass-clauses
where this a grammatical option. In other words, compared to the native speakers,
learners typically overuse the following sequences:
[a2017_03] Ich denke, dass es zu viel Werbung gibt aber wenn die Jugend... ‟I think that there is too much advertising but if young people…‟
122
This is further underlined by the fact that some of these expressions are even further modified by adverbs like persönlich (perosonally) and selbst (myself) as in: „Ich persönlich/selbst glaube…‟ (I personally/myself think…)
191
[b1034_01] Ich glaube, dass Situationen wie dieser selten sind, aber... ‟I believe that situations like these are rare, but…‟
[c1055_04] Ich bin der Meinung, dass Gewalt in den Medien eine große Rolle spielt. ‟I am of the opinion that violence in the media plays a significant role.‟
[c2036_01] Meiner Meinung nach ist die Einführung von Studiengebühren prinzipiell sinnvoll, ... ‟In my opinion, the introduction of tuition fees is, in principle, a sensible thing, …‟
The native speakers, on the other hand, make more use of the full breadth of
available expressions (while at the same time using these items less altogether123) and
employ them in a more varied way, alternating between sentence-initial and sentence-
internal positions (for clausal and prepositional expressions) and connecting clauses with
and without the conjunction dass (that), e.g.:
[keds_048] Dieses vom Autor ebenfalls nicht weiter kommentierte Argument ist meiner Ansicht nach zutreffend. „This argument – which is also not further commented on by the author – is, in my opinion, correct.
[keds_035] Das bestehende Risiko kann, denke ich, noch weiter abgesenkt werden, wenn diese Wissenschaftler nicht Hals über Kopf anfangen würden am Menschen zu experimentieren. „The existing risk can, I think, be even further reduced if scientist would not be in a mad rush to start experimenting on human beings.‟
[keds_037] Ich denke, wenn man nur einige Regeln des sanften Tourismus einhält ist ein Urlaub für beide Seiten - Urlauber und Einwohner – vorteilhaft. „I think if only some rules of sustainable tourism are observed then a holiday is beneficial for both sides – holiday-makers and natives.‟
Another interesting fact becomes apparent if we look back at the results for modal
verbs, namely that the overuse of periphrastic modal expressions in this category goes
hand in hand with the underuse of the corresponding modal verb DÜRFTE (which is one
of the modal verbs that do not have a form/meaning equivalent in English). In section
5.1.1.2 it was established that this modal verb was not part of the learners‟ acquired
repertoire of modal means. We could speculate that the fact that there is a variety of
periphrastic modal expressions for this particular modal meaning might be a factor in the
non-acquisition of this modal verb as the semantic gap is easily filled with adverbial,
verbal or nominal phrases.
123
They do, however, use them considerably more than expert writers, as the count of these items in LIMAS reveals a negligible frequency of just 0.6 per 10,000 words. We will discuss the implications of this result in Chapter 6.
192
The same goes for the category „tentative conclusion‟. It was shown in 5.1.1.3 that the
corresponding epistemic modal verb meanings SOLLTE and MÜSSTE have not been
acquired by learners even in Year C. PMEs that express similar meanings, on the other
hand, are used frequently by the learners. The data gives the following detailed picture:
Year A Year B Year C KEDS
scheinen 13
(2.8) scheinen
50 (7.6)
scheinen 48
(5.5) wohl
46 (5.6)
normalerweise 12
(2.6) normalerweise
21 (3.2)
wahrscheinlich 17
(1.9) scheinen
11 (1.3)
wahrscheinlich 6
(1.3) wahrscheinlich
18 (2.7)
normalerweise 16
(1.8) wahrscheinlich
11 (1.3)
wohl 1
(0.2) vermutlich
3 (0.5)
wohl 12
(1.4) scheinbar
7 (0.9)
annehmen 2
(0.3) offenbar
8 (0.9)
anscheinend 7
(0.9)
scheinbar 1
(0.2) höchst-wahrscheinlich
4 (0.5)
offenbar 6
(0.7)
offenbar 1
(0.2) scheinbar
2 (0.2)
normalerweise 3
(0.4)
anscheinend 1
(0.2) anscheinend
1 (0.1)
ist anzunehmen 1
(0.1)
annehmen 1
(0.1) schätzungsweise
1 (0.1)
TOTAL 32** (6.9)
TOTAL 97
(14.7) TOTAL
109 (12.4)
TOTAL 94
(11.5)
TABLE 51: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING
TENTATIVE CONCLUSIONS (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
As with other categories that comprise a wide variety of items we can see a marked
increase in the number of different types of items that occur between Year A and Year B.
This broadening in lexical variety from Year A to Year B also coincides with an
increasing frequency overall for this category, essentially turning the significant underuse
in Year A into overuse in Year B.
In this category the first item alone makes up more than 40% of all occurrences, going
up to as much as 52% in Year B (Year A: 40.6%; Year B: 52%; Year C: 44%; KEDS:
49%). Curiously, this first item is the same in all learner groups, scheinen (seem), but a
different one for KEDS: wohl (probably). Scheinen is ranked second highest in KEDS,
but with a much lower frequency than in the learner groups. And once again, there are
193
distinct qualitative differences in the colligational patterns of this verb between the
learners and native speakers.
In the learner data, the prominent sequence is with the dummy-subject es (it):
[a1018_02] Es scheint, dass das sogenannte Denglish für die jungere Generationen internationaler und einfacher zu benutzen ist. ‟It seems that so-called Denglish is more international and easier to use for the younger generation.‟
[c1046_01] Es scheint als ob, Täter die Strafen nicht mehr fürchten. ‟It seems as though delinquents are not afraid of punishment anymore.‟
This sequence constitutes up to 92% of the occurrences of the verb scheinen in the
learner data, but fewer than 10% in KEDS. It is positioned sentence-initial in 50% of
cases in all three learner groups:
Year A Year B Year C KEDS
sequence: es scheint 12 32 33 1
as % of all instances of scheinen 92.3 64.0 68.8 9.1
sentence-initial es scheint 6 16 16 1
as % of all instances of es scheint 50.0 50.0 48.5 100.0
TABLE 52: COLLIGATIONAL SEQUENCE ES SCHEINT
Furthermore, in the learner data this sequence is often extended by the 1st person
personal pronoun in the dative case mir to the phrase Es scheint mir…, where the
dummy-subject can also be deleted: Mir scheint…. These sequences are prominent in
Years A and B, less in Year C and do not occur in KEDS.
Year A Year B Year C KEDS
sequence: mir scheint/es scheint mir 7 10 4 0
as % of all instances of scheinen 53.8 20.0 8.3 0.0
TABLE 53: COLLIGATIONAL SEQUENCE MIR SCHEINT/ES SCHEINT MIR
In Year A, this sequence is often even further extended into Mir scheint es so, dass… (5
instances), which does not occur in the native speaker data. From Year B, this is
replaced by the more appropriate Es scheint, als ob… (6 instances; Year C: 9; KEDS: 1).
The two modal adverbs that are ranked second and third in all learner groups are
normalerweise (usually) and wahrscheinlich (probably), respectively. The latter is also
ranked third in KEDS, the former, however, only occurs three times (rank 7). For both
194
adverbs, no unusual patterns could be detected. There are occasional word order errors
with normalerweise in Year A despite the relative freedom in sentence position, but these
have practically disappeared by Year B.
[a2013_03] Es ist ein groß Problem für die Eltern, weil *normalerweise man achtzehn sein muss, um pornografische Material zu kaufen, ... (target: ... weil man normalerweise achtzehn sein muss...) „It is a big problem for the parents, because usually you have to be eighteen to be able to buy pornographic material …‟
Wahrscheinlich does not seem to pose a problem where word order is concerned. Also
in terms of meaning, the learners – with a few exceptions – generally employ both items
in appropriate contexts, e.g:
[b1022_07] Die Propaganda, die am häufigsten benutzt wird ist Werbung, obwohl sie normalerweise nicht als Propaganda genannt werden. ‟The type of propaganda that is used the most is advertising, although it is usually not called propaganda.‟
[b1009_05] Wahrscheinlich arbeiten wir deshalb so viel, weil unsere Freizeit so viel kostet; ‟We probably work so much, because our leisure time is so expensive;‟
The top item in the native speaker data – wohl – does not feature much at all in Years
A and B. This is an adverb that has been acquired by some learners in Year C, although
its frequency is much lower than in KEDS. In addition, 7 out of the 12 instances of wohl
in Year C stem from the same author. Despite its correct use in these instances, it is
clearly only the preference of one individual learner for this item that pushes it up to
fourth rank in Year C, and it cannot be considered part of the generally acquired
repertoire of PMEs.
The next category – PMEs denoting the assumption of a possibility (where the
corresponding modal verbs are KÖNNEN and MÖGEN) – is a category that is
characterised by an increasing overuse on the side of the learners from Year A to Year
C:
195
Year A Year B Year C KEDS
vielleicht 35
(7.6) vielleicht
84 (12.8)
vielleicht 134
(15.3) vielleicht
48 (5.8)
möglich/ Möglichkeit
11 (2.4)
möglich/ Möglichkeit
15 (2.3)
möglich/ Möglichkeit
22 (2.5)
denkbar 10
(1.2)
denkbar 1
(0.2) möglicher- weise
4 (0.6)
möglicher- weise
6 (0.7)
möglicher- weise
9 (1.1)
möglicher- weise
1 (0.2)
denkbar 1
(0.2) eventuell
2 (0.2)
möglich/ Möglichkeit
7 (0.9)
Chance 1
(0.2) Chance
1 (0.2)
unter Umständen
1 (0.1)
eventuell 7
(0.9)
eventuell 1
(0.2)
vorstellbar
1 (0.1)
unter Umständen
1 (0.2)
unter Umständen
1 (0.1)
TOTAL 48
(10.4) TOTAL
107*** (16.2)
TOTAL 165*** (18.8)
TOTAL 83
(10.1)
TABLE 54: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING
ASSUMPTIONS OF POSSIBILITY (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
As with the last category, we can see that one item dominates in terms of frequency.
Unlike in the previous category, however, here this is the same one in all learner and
native speaker corpora, namely the adverb vielleicht (perhaps). It is already a firm part of
the learners‟ repertoire in Year A and, proportionally, it is not overused much in this
group compared to KEDS. The item starts being overused in Year B and reaches a level
in Year C that is almost three times as high as the native speakers. In terms of both
meaning and sentence positioning no diverging patterns could be detected between
learners and native speakers, i.e. the learners use vielleicht in the same way as the
native speakers, they simply use it much more often124.
[c1046_04] Es ist für Journalisten vielleicht schwer, die Konsequenzen ihrer Berichterstattung abzuschätzen. „It is perhaps difficult for journalists to assess the consequences of their reports.‟
[b1031_02] Immer mehr Väter wollen einen großen Einfluß in dem Leben ihrer Kinder und vielleicht ist die Situation nicht so hoffnungslos für Frauen als in der Vergangenheit.
124
A search in the general German language corpus LIMAS reveals that vielleicht (perhaps) only occurs with 2.8 instances per 10,000 words, which means that this item is also used more in the specific text type of KEDS. Furthermore, a search of the English equivalent „perhaps‟ in LOCNESS comes up with a frequency of 3.4 per 10,000 words; the overuse on the part of the learners can therefore not be explained by a more frequent use of the equivalent L1 item.
196
„More and more fathers want to have a big influence on the lives of their children and perhaps the situation is not as hopeless for women as in the past.‟
One differentiating factor between learner and native speaker use of vielleicht that could
be discovered was the propensity for it to co-occur with modal verbs and the subjunctive
mood in the learner data more than in KEDS.
The second most frequent item in the learner data is patterns with the adjective
möglich (possible) and the noun Möglichkeit (possibility), respectively. They are
distributed as follows:
Year A Year B Year C KEDS
adjective: möglich 9 10 9 2
noun: Möglichkeit 2 5 13 5
TABLE 55: DISTRIBUTION OF ADJECTIVE MÖGLICH AND NOUN MÖGLICHKEIT
As in previous categories, we find the familiar sequence with the dummy-subject es as
the prevailing pattern in the learner groups. For the adjective this results in the sequence
es ist möglich… (it is possible…) and for the noun in the sequence es gibt die
Möglichkeit… (There is the possibility…):
Year A Year B Year C KEDS
sequence: es ist möglich 8 7125
7 2
sequence: es gibt die Möglichkeit 1 1 5 0
TABLE 56: COLLIGATIONAL SEQUENCES ES IST MÖGLICH/ES GIBT DIE MÖGLICHKEIT
This sequence is furthermore mostly followed by a subordinate dass-clause, so that the
following patterns emerge:
[c1054_04] Wenn sie eine Sendung mit Gewalt sehen denn ist es möglich, dass sie von dieser Gewalt beeinflusst werden. „When they watch a programme with violence then it is possible that they are influenced by this violence.‟
[c2039_03] In der Tat gibt es immer die Möglichkeit, dass die deutsche Geschichte sich widerholen könnte. „There is, indeed, always the possibility that the German history could
125
This includes two variations, namely “Es sollte möglich sein...“ (It should be possible...) and “Es wäre möglich“ (It would be possible).
197
repeat itself.‟
The category that represents expressions for evidential meanings (modal verbs
SOLLEN and WOLLEN) does not feature frequently in either learner or native speaker
subcorpora.
Year A Year B Year C KEDS
angeblich 2
(0.3) angeblich
13 (1.5)
angeblich 6
(0.7)
gerüchteweise 1
(0.1)
TOTAL 0 TOTAL 2
(0.3) TOTAL
13 (1.5)
TOTAL 7
(0.9)
TABLE 57: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING
EVIDENTIAL MEANING (SOLLEN/WOLLEN)
From these figures, it seems like the adverb angeblich (allegedly) has been acquired by
learners in Year C. A closer look at the concordances reveals, however, that the 13
occurrences there originate from only 5 different authors (one learner using angeblich 5
times in a single text) and, furthermore, that 5 out of these 13 instances are actually
incorrect uses. This means, essentially, that only two learners have acquired the adverb
correctly, the other three learners have not grasped the exact meaning of it yet, as the
following sentences exemplify:
[c1050_02] Einige Leute aber glauben, dass es wirklich die Verantwortung der Eltern ist und deshalb würden Lehrer den Beruf als „Eltern-Ersatz“ haben. Sie haben Angst, dass in der Zukunft die Sanktionsmöglichkeiten immer schlimmer *angeblich werden. „But some people believe that it is really the responsibility of the parent and therefore teachers would have the job of “parent replacement”. They are afraid that in the future the disciplinary options will *allegedly become worse and worse.‟
[c1047_02] Laut dem Artikel muss gefragt werden, warum andere Religionen, zum Beispiel Christentum *angeblich bevorzugt werden. „According to the article it must be asked why other religions, e.g. Christianity, are *allegedly being privileged.‟
5.2.2.3 Negative periphrastic modal expressions
The final group of PMEs to be investigated in this section on epistemic expressions are
the two categories to do with negativity: PMEs that achieve a hedging of negations and
198
those that intensify negations. As the following chart shows, both of these categories are
(mostly significantly) underused by learners in all three year groups, despite increasing
frequencies from Year A through to Year C:
FIGURE 17: FREQUENCIES OF PMES AFFECTING NEGATION
In the first of these categories (hedging negations) the low frequencies in the learner
data stems entirely from the considerable underuse of one item: kaum (hardly):
Year A Year B Year C KEDS
unwahrschein-lich
1 (0.2)
kaum 3
(0.5) kaum
5 (0.6)
kaum 42
(5.1)
unwahrschein-lich
1 (0.2)
unwahrschein-lich
4 (0.5)
unwahrschein-lich
3 (0.4)
Zweifel 3
(0.3) fast
2 (0.2)
scheinbar 1
(0.1)
TOTAL 1*** (0.2)
TOTAL 4*** (0.6)
TOTAL 12*** (1.5)
TOTAL 48
(5.8)
TABLE 58: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES ACHIEVING A
HEDGING OF NEGATIONS (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
Year A Year B Year C KEDS
freq. per
10,000 words
hedging negation
intensifying negation
199
Whereas the learners hardly make use of it, the native speakers employ this item
frequently in order to attenuate a statement:
[keds024] Seine statistischen Beispiele zeigen auch , dass Gewalt alltäglich ist und kaum noch als außergewöhnlich angesehen wird. ‟His statistical examples also show that violence is commonplace and is hardly seen as extra-ordinary anymore.‟
The underuse of this „negative‟ hedging category matches the underuse of the category
that contains items that result in positive hedged propositions (see 5.2.2.1 above). These
two categories seem to indicate that learners avoid items that result in a toning down of
either a categorical assertion or negation.
The items in the second category in this group achieve the opposite, namely produce
an intensifying effect on the negation.
Year A Year B Year C KEDS
nie 13
(2.8) nie
16 (2.4)
nie 23
(2.6) nie
24 (2.9)
überhaupt nicht
3 (0.5)
gar nicht 10
(1.1) gar nicht
16 (1.9)
auf keinen Fall
2 (0.3)
auf keinen Fall 6
(0.7) auf keinen Fall
6 (0.7)
niemals
1 (0.2)
überhaupt nicht 4
(0.5) überhaupt nicht
5 (0.6)
niemals
2 (0.2)
niemals 4
(0.5)
keineswegs
4 (0.5)
ausgeschlossen
2 (0.2)
TOTAL 13*** (2.8)
TOTAL 22*** (3.3)
TOTAL 45
(5.2) TOTAL
63 (7.5)
TABLE 59: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES ACHIEVING AN
INTENSIFICATION OF NEGATIONS (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
We can see in this category how the learners‟ active vocabulary use gradually expands
from Year A to Year C, incorporating more and more items until the list resembles that of
the native speakers. Once they are acquired, the learners do not seem to have any
difficulties with these items and no learner-specific patterns could be found.
200
[a2021_02] 'Fremdsprache' Filme sind oft bei Kritikern erfolgreich, aber nie unter der englischsprechenden Bevölkerung. ‟‟Foreign language films‟ are often successful with critics, but never with the English-speaking public.‟
[c1047_01] Auf keinen Fall kann einen Mord mit einem anderen Mord rechtfertigt werden, denn der Scharfrichter wäre genauso schuldig wie der Mörder selbst. ‟Under no circumstances can one murder be justified by another murder, because the executioner would be as guilty as the murderer himself.‟
It is worth noting, however, that the significant underuse of intensifying negative PMEs in
Years A and B is congruent to a significant underuse of purely assertive PMEs, which
achieve an intensifying effect on positive assertions.
5.2.3 Non-epistemic periphrastic modal expressions
From the overview chart (Figure 13; p. 172) we have seen that non-epistemic PMEs are
much less frequent than epistemic ones. As mentioned before, this corresponds with
high frequencies of non-epistemic modal verbs. The following chart gives an overview of
non-epistemic PMEs:
201
FIGURE 18: FREQUENCIES OF NON-EPISTEMIC PMES
For epistemic modality we have seen that there seems to be some interrelation between
the non-acquisition of certain modal verbs and high frequencies of the corresponding
PMEs. Where non-epistemic modality is concerned, no such interrelations can be found
as all modal verb meanings have been confidently acquired before Year A126. It does
appear, however, that there is a tendency for learners to underuse those PME categories
where the corresponding modal verb has an English form/meaning equivalent: MÜSSEN
(duty/compulsion), KÖNNEN (possibility/ability) and SOLLEN/SOLLTE127 (instruction and
suggestion/advice), while PMEs that correspond to modal verbs without English
form/meaning equivalents tend to be overused: WOLLEN (volition), MÖCHTE
(attenuated volition), MÖGEN (affection/liking) and DÜRFEN (permission). This effect is
most pronounced in Year A, but still visible in Years B and C. One explanation for this
could be that the learners do not see the need for alternative modal expressions as they
126
This is with the exception of the subjunctive modal verb forms (especially the difference between SOLLEN and SOLLTE). The learners, however, do not seem to perceive this as a problem and certainly do not avoid these verbs.
127 Strictly speaking, SOLLEN does not have an English form/meaning equivalent, but it is included here as we have seen that the learners often are not able to differentiate between the meaning of SOLLEN and SOLLTE.
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
18.0
freq. per
10,000 words
Year A
Year B
Year C
KEDS
duty
compulsion
possibilityabilitysuggestion
advice
instructionvolitionattenuated
volition
affection
liking
permission
202
feel confident in the use of the modal verbs for these particular non-epistemic meanings.
Conversely, where the learners are used to expressing certain notions by PMEs in their
L1 rather than by modal verbs, they tend to follow this pattern in the L2 and overuse
PMEs, despite having acquired the L2 specific modal verb meanings. The following
sections will explore how individual items are used within the different categories. For the
sake of completeness each category and all items will be displayed, but as there are
several categories with very low frequencies, only the bigger categories will be discussed
in more detail.
The first category includes PMEs that express non-epistemic necessity, i.e. duty or
compulsion (corresponding to the modal verb MÜSSEN):
Year A Year B Year C KEDS
notwendig/ Notwendigkeit
4 (0.9)
notwendig/ Notwendigkeit
9 (1.4)
nötig 12
(1.4) nötig
12 (1.5)
erforderlich 2
(0.4) nötig
4 (0.6)
notwendig/ Notwendigkeit
9 (1.0)
notwendig/ Notwendigkeit
10 (1.2)
nichts anderes übrig
1 (0.2)
gezwungen 3
(0.5) verpflichtet
5 (0.6)
erforderlich 4
(0.5)
nicht nötig 1
(0.2) erforderlich
2 (0.3)
gezwungen 5
(0.6) gezwungen
2 (0.2)
unerlässlich
1 (0.2)
erforderlich 2
(0.2) verpflichtet
2 (0.2)
unvermeidlich
1 (0.2)
keine Wahl 2
(0.2) unerlässlich
1 (0.1)
verpflichtet
1 (0.2)
unvermeidlich 2
(0.2) unumgänglich
1 (0.1)
keine Wahl
1 (0.2)
TOTAL 8*
(1.7) TOTAL
22 (3.3)
TOTAL 37
(4.2) TOTAL
32 (3.9)
TABLE 60: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING DUTY OR
COMPULSION (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
We can see that already in Year B the learners have broadened their vocabulary range
in this category to native speaker standard. Another noteworthy point here is that the
frequencies are spread more evenly across the different items than in many of the
epistemic categories, and a look at the concordances confirms that it is not just one or
two individual learners who use a greater variety of items but that the use is distributed
203
more widely across learners. On top of this, no distinct patterns, sequences or
collocations could be found, although items like notwendig and nötig (necessary) would
lend themselves to the same kind of sequences that we have seen with möglich in
epistemic modality, i.e. Es ist notwendig/nötig... (It is necessary…). However, the
learners in all year groups employ these items in a range of different sentence
constructions.
[a2001_03] Wenn sie älter sind und wollen gute Arbeiten, werden Benutzung von dem Internet und moderne Technologie notwendig sein. „When they are older and want a good job using the internet and modern technology will be necessary.‟
[b1003_05] Die Franzosen vertreten für die 35 Stunden Woche, sie bestehen, dass es keine Notwendigkeit gibt zu arbeiten mehr. „The French defend the 35-hour working week, they insist that there is no necessity to work more.‟
[c2039_01] Außerdem sind sie verpflichtet den Arbeitgeber und Versicherer es zu sagen. „Furthermore they are required to inform their insurance company about it.‟
PMEs that denote instructions/obligations (modal verb SOLLEN) or
suggenstions/advice (modal verb SOLLTE) hardly occur at all in any of the subcorpora:
Year A Year B Year C KEDS
verlangt 1
(0.2) verlangt
2 (0.3)
verlangt 2
(0.2) verlangt
5 (0.6)
geboten 1
(0.1)
TOTAL 1
(0.2) TOTAL
2 (0.3)
TOTAL 2
(0.2) TOTAL
6 (0.7)
TABLE 61: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING
INSTRUCTIONS/OBLIGATIONS
Year A Year B Year C KEDS
wäre gut 1
(0.2) wäre nicht richtig
1 (0.1)
wäre angebracht 1
wäre ratsam 1
TOTAL 1
(0.2) TOTAL TOTAL
1 (0.1)
TOTAL 2
(0.2)
TABLE 62: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING
SUGGESTIONS/ADVICE
204
The same is true for PMEs denoting volition:
Year A Year B Year C KEDS
vorhaben 2
(0.4) Plan
2 (0.3)
Wille 2
(0.2) vorhaben
2 (0.2)
beabsichtigen 1
(0.2 keine Lust
2 (0.3)
beabsichtigen 1
(0.1) auf…aus sein
1 (0.1)
beabsichtigen 1
(0.2) vorhaben
1 (0.1)
Wille 1
(0.1)
Wille 1
(0.2)
TOTAL 1
(0.2) TOTAL
2 (0.3)
TOTAL 2
(0.2) TOTAL
6 (0.7)
TABLE 63: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING VOLITION
In the few instances that the learners use any of the items from these last three
categories, they generally do not have any problems with their semantic meaning or
grammatical requirements.
The next category that is worth looking into in more detail is PMEs denoting
attenuated volition or wishes (modal verb MÖCHTE):
Year A Year B Year C KEDS
hoffen 7
(1.5) hoffentlich
17 (2.6)
hoffen 31
(3.5) Hoffnung
6 (0.7)
hoffentlich 5
(1.1) hoffen
15 (2.3)
hoffentlich 12
(1.4) hoffen
4 (0.5)
wünschen 2
(0.4) Wunsch
5 (0.8)
Wunsch 9
(1.0) wünschen
4 (0.5)
würden lieber 1
(0.2) würde lieber
5 (0.6)
Wunsch 3
(0.4)
hätten gern 1
(0.2) wünschen
3 (0.3)
keine Lust 2
(0.2)
TOTAL 14
(3.0) TOTAL
39*** (5.9)
TOTAL 62*** (7.1)
TOTAL 17
(2.1)
TABLE 64: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING
ATTENUATED VOLITION/WISHES (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
205
In the learner data, the verb hoffen (hope) and the adverb hoffentlich (hopefully) are the
two most frequent items. Interestingly, hoffentlich does not occur at all in the native
speaker data. This can possibly be explained with a greater tendency of the native
speakers to avoid items that inherently express speaker involvement, which was already
shown for the epistemic PMEs (cf. 5.2.2.2). This also explains to some extent the high
frequency of hoffentlich in Year B, as this group contains a portion of texts (year abroad
projects) where the task requires the students to reflect upon their own hopes, fears and
wishes for their imminent year in Germany. On top of that the verb hoffen is exclusively
used with first person in Year B, which again can be assigned to the nature of that
particular task (26 out of overall 32 instances – 81% – of hoffen and hoffentlich in Year B
occur in the year abroad projects).
In Year C, however, the extreme prominence of the verb hoffen is due to the overuse
of a different pattern: Es bleibt zu hoffen, dass… (It remains to be hoped that…):
[c2045_06] Es bleibt zu hoffen, dass die Grauzone der aktiven Sterbehilfe aufgeklärt wird. „It remains to be hoped that the grey area surrounding euthanasia is going to be clarified.‟
[c1058_04] Es bleibt zu hoffen, dass mehr Leute zur Vernunft kommen und den Vorschlag unterstützten werden. „It remains to be hoped that more people see sense and support the suggestion.‟
This sequence occurs exclusively in this year group, where it makes up 24 of the 31
instances of hoffen (77%). Out of these, 21 are placed in a thematising sentence-initial
position, typically towards the end of the text. This indicates that the learners in Year C
strive towards more neutral expressions with respect to speaker involvement, but then
tend to use this same sequence over and over again. Incidentally, this category in this
year group is the only one where the PME frequency is higher than the corresponding
modal verb category in the learner data.
For KEDS the only category where PMEs are more frequent than modal verbs is the
one denoting general affection or liking (MÖGEN).
Year A Year B Year C KEDS
lieben 3
(0.7) lieben
4 (0.6)
lieben 4
(0.4) neigen zu
6 (0.7)
gern 2
(0.4) nicht leiden können
2 (0.3)
gern haben 1
(0.1) gern
1 (0.1)
206
nicht Lust 1
(0.2) gern
1 (0.2)
neigen zu 1
(0.1) ablehnen
1 (0.1)
ablehnen 1
(0.2) neigen zu
1 (0.2)
nicht leiden können
1 (0.1)
hassen 1
(0.2)
TOTAL 7
(1.5) TOTAL
9 (1.4)
TOTAL 6
(0.7) TOTAL
9 (1.1)
TABLE 65: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING
GENERAL AFFECTION OR LIKING
As the overall frequencies for this category of PMEs is very low for learners and
native speakers, we shall not discuss it in great detail. One point that is interesting to
note, however, is that the top item in all learner groups – the verb lieben (love) – does
not occur at all in KEDS. Here, the verb neigen zu (be prone to do sth./to tend toward
sth.) is the most frequent item. The difference between these two verbs is that lieben is
primed for positive contexts, i.e. what is „being loved‟ is usually something pleasant,
whereas neigen zu is primed for undesirable behaviour. So, the learners use the PMEs
in this category to describe circumstances that they view as positive (examples from the
corpora are: German films, Germany, vegetables, work, time with the family, football
etc.):
[b1006_05] Es gibt auch viele Leute, die ihren Arbeiten lieben. Sie wecken jeden Morgen auf und denken, dass sie sehr glücklicher ihrer Arbeit zu haben sind. „There are also many people who love their work. They wake up every morning and think that they are very lucky to have their work.‟
[a2008_03] Ich liebe besonders deutschen Filmen weil sie immer verrückt sind. „I particularly love German films because they are always crazy.‟
The native speakers, on the other hand, tend to use neigen zu and associate negative,
undesirable behaviour with it128 (examples from KEDS are: selfishness, aggression,
violence, lynch law):
[keds_006] Allerdings kann ein solches Computerspiel Anleitung zum Schlagen sein, wenn der Jugendliche sowieso schon dazu neigt, Gewalt anzuwenden. „Such a computer game can, however, be a guide on how to beat people, if the young person is prone to violent behaviour anyway.‟
128
In the two instances that occur in the learner data, the one in Year B does not follow this priming, the one in Year C does.
207
[keds_038] Dies zeigte sich bei den verschiedenen „Kinderschänderprozessen“ der letzten Zeit, bei denen die Öffentlichkeit zu einer Lynchjustiz der Angeklagten neigte, noch bevor deren Schuld festgestellt wurde. „This became apparent in the various recent “child abuse cases”, where the public tended towards lynch law for the accused before their guilt was even established.‟
In the corresponding modal verb category MÖGEN (Section 5.1.2.7) it could be
established that the native speakers tend to refrain from the kinds of statements that
express positive likings of the subjects and the data from the PMEs seems to support
this trend. The learners, particularly in Years A and B, overuse the modal verb as well as
the corresponding PMEs to describe exactly those circumstances.
The next two categories to be discussed are concerned with PMEs denoting
possibility – either in terms of circumstances that allow for the proposed action to take
place or in terms of personal abilities (modal verb KÖNNEN). The former one is by far
the more frequent category although the lexical range is more limited. It only comprises
the adjective möglich (possible) and the nouns Möglichkeit (possibility) and Chance
(opportunity). On top of this, however, adjectives derived in a productive morphological
process from a verb stem plus suffix –bar also belong to this category (cf. Table 67). As
these items often only occur once but occasionally occur more often, the following table
indicates the number of unique types of –bar-adjectives in each subcorpus as well as
their overall frequency.
Year A Year B Year C KEDS
(un)möglich/Möglichkeit
23 (5.0)
(un)möglich/ Möglichkeit
41 (6.2)
(un)möglich/ Möglichkeit
78 (8.9)
(un)möglich/ Möglichkeit
99 (12.1)
Chance 3
(0.7) *bar (9 types)
9 (1.4)
*bar (15 types) 21
(2.4) *bar (28 types)
42 (5.1)
Chance 14
(1.6) Chance
5 (0.6)
TOTAL 26*** (5.6)
TOTAL 50*** (7.6)
TOTAL 113** (12.8)
TOTAL 146
(17.8)
TABLE 66: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING
POSSIBILITY DUE TO CIRCUMSTANCES (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
The table shows that there is a steady increase in the number of –bar-adjectives that
occur in the learner data – from 0 in Year A to 15 in Year C, although even this figure is
208
only about half of that in KEDS. The following list gives the different types that are found
in each subcorpus:
Year B Year C KEDS
anstellbar bemerkbar abbaubar
bemerkbar beweisbar absehbar
*erhältbar (target: erhältlich) erkennbar anwendbar
unannehmbar erreichbar beeinflussbar
unberechenbar finanzierbar befahrbar
ungenießbar heilbar begründbar
vermeidbar unberechenbar behandelbar
voraussagbar unbezahlbar brauchbar
vorhersehbar unerreichbar durchschaubar
unheilbar durchsetzbar
unvorstellbar einhaltbar
vereinbar erkennbar
verfügbar erreichbar
vergleichbar finanzierbar
*verstandbar (target: verständlich)
haltbar
kontrollierbar
nachvollziehbar
realisierbar
rückgängigmachbar
übersehbar
unbesiegbar
unpassierbar
unvorhersehbar
verfügbar
vergleichbar
vermeidbar
vertretbar
voraussehbar
TABLE 67: TYPES OF –BAR-ADJECTIVES IN THE PME CATEGORY „POSSIBITLITY DUE TO
CIRCUMSTANCES‟ (ADJECTIVES THAT OCCUR IN CLEG AND KEDS IN BOLD)
The list shows that many of the items, although they are not new word creations
altogether, are arrived at by way of some creative derivational process, which is
apparent in the two errors that occur in the learner data where this process unfortunately
produces a non-target like form.
209
With regard to the most frequent item in this category (möglich/Möglichkeit), it is, once
again, the familiar pattern with the dummy-subject es that prevails in combination with
the adjective (un)möglich:
[a2003_01] Es ist auch möglich im Internet Musik zu herunterladen und das ist illegal und manchmal kennen Kinder das nicht, ... „It is also possible to download music from the internet and that is illegal and sometimes children don‟t know that, …‟
[b1031_01] Jetzt ist es möglich eine weite Reihe von Zeitungen zu kaufen. „It is now possible to buy a wide range of newspapers.‟
[c1054_04] Also ist es fast unmöglich diese Gewalt zu vermeiden. „So it is almost impossible to avoid this kind of violence.‟
This colligational pattern makes up a stable proportion of around 50% of occurrences of
this adjective in the learner data, but only a much smaller proportion in KEDS:
Year A Year B Year C KEDS
sequence: es ist (un)möglich... 10 14 23 9
as % of all instances of (un)möglich 52.6 43.8 48.9 18.3
TABLE 68: COLLIGATIONAL SEQUENCE ES IST (UN)MÖGLICH…
A further noteworthy point is that we can see clearly how the colligation pattern here
resolves the seeming ambiguity between epistemic and non-epistemic uses of the
adjective möglich. Epistemic möglich colligates with a subordinate dass-clause while
non-epistemic möglich colligates with an infinitive clause (with infinitive marker zu). The
same distinction is true for the English adjective „possible‟. It is therefore not surprising
that the learners do not have any problems with these two constructions.
As for the noun Möglichkeit, there are three verb collocates that occur in differing
proportions in the learner groups compared to KEDS. In KEDS the most frequent one is
haben (have) (11 out of 49 instances), as in
[keds_30] Durch das Internet hat jeder die Möglichkeit sich und seine Gefühle und Gedanken der ganzen Welt zu offenbaren. „With the internet everybody has the opportunity to lay open their feelings and thoughts for the whole world to read.‟
210
In the learner data there is a shift from Year A, where the most frequent verb collocate is
geben in the sequence es gibt die Möglichkeit...129 (3 out of 4 instances) to Years B and
C, where the most frequent verb collocate is haben (4 out of 9 and 11 out of 31
instances, respectively) as in the native speaker data:
[a2017_02] Trotz der Nachteile gibt es viele Vorteile und auch heutzutage gibt es eine Möglichkeit die Benutzung des Internet zu kontrollieren. „Despite the disadvantages there are many advantages and nowadays there is a possibility of controlling the use of the internet.‟
[b1017_05] Und doch, mussen wir darauf hinweisen, dass wir die möglichkeit neue Leute kennen zu lernen und andere Erfahrungen zu sammeln haben, wenn man eine Stelle hat. „And still we have to point out that we have the opportunity to meet new people and have different experiences if you have a job.‟
[c2044_04] Während "Goodbye Lenin" die Unterschiede zwischen Ost und West deutlich darstellte, hat die Medien auch die Möglichkeit diese Unterschiede zu minimieren. „While “Goodbye Lenin” clearly portrayed the differences between East and West, the media also has the opportunity to minimise these differences.‟
In comparison to the corresponding modal verb KÖNNEN, it is noted that PMEs in this
category are significantly underused by all learner groups, while the modal verb is
overused in Year A and significantly underused in Years B and C. So, apart from Year A
there do not seem to be any compensatory factors at work between modal verb and
PME use, i.e. no overuse of modal verbs at the expense of PMEs or vice versa.
The other category that corresponds to uses of the modal verb KÖNNEN – denoting
possibility due to personal abilities – is also underused consistently by the learners:
129
Incidentally resulting in the same sequence as for epistemic modality.
211
Year A Year B Year C KEDS
Fähigkeit 2
(0.4) fähig/ Fähigkeit
2 (0.3)
fähig/Fähigkeit 6
(0.7) in der Lage
17 (2.1)
Erfahrung 1
(0.2) in der Lage
5 (0.6)
fähig/Fähigkeit 8
(1.0)
Kenntnis 1
(0.2)
unmöglich/ keine Möglichkeit
4 (0.5)
Erfahrung 1
(0.1)
nicht beherrschen
1 (0.1)
nicht möglich 1
(0.1)
TOTAL 2*** (0.4)
TOTAL 4*** (0.6)
TOTAL 17
(1.8) TOTAL
27 (3.3)
TABLE 69: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING
POSSIBILITY DUE TO PERSONAL ABILITIES (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
In addition to that, at least one of the two instances in Year A is actually incorrect –
the noun Fähigkeit (ability) is used in a context where only Möglichkeit (possibility) is
correct, as what is being described is a possibility due to school regulations, not the
personal ability of the pupil:
[a2011_03] Obwohl es die *Fähigkeit (target: Möglichkeit), von eine Hauptschule oder Realschule bis einem Gymnasium zu wechseln gibt, ist es nicht so allgemein. „Although there is the *ability (target: possibility) to change from Hauptschule to Realschule or even Gymnasium it is not that common.‟
Other than that, no diverging patterns could be detected in the learner data. In Years B
and C the expressions are all used in appropriate contexts.
Finally, the last category to be discussed relates to PMEs that denote permission
(modal verbs DÜRFEN and KÖNNEN). Here we can see underuse in Years A and B and
a significant overuse in Year C compared to KEDS:
212
Year A Year B Year C KEDS
erlaubt/ Erlaubnis
6 (1.3)
erlaubt 4
(0.6) verboten
31 (3.5)
verboten 13
(1.6)
verboten 1
(0.2) verboten
4 (0.6)
erlaubt/ Erlaubnis
23 (2.6)
erlaubt 5
(0.6)
gestattet 2
(0.2)
TOTAL 7
(1.5) TOTAL
8 (1.2)
TOTAL 54*** (6.2)
TOTAL 20
(2.4)
TABLE 70: RAW FREQUENCIES (FREQ. PER 10,000 WORDS IN BRACKETS) OF PMES DENOTING
PERMISSION (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
This is the only category where – due to the participle verboten (prohibited) – the
negative version of the category (NICHT DÜRFEN) is more frequent than the positive.
No learner specific patterns could be detected with this participle. However, when
looking at the participle with opposite meaning – erlaubt (allowed) – we see that the
learners in all year groups struggle with the grammatical structure that this participle
requires. In the English construction „X is allowed to do Y‟, the person being permitted to
do something is the grammatical subject of the sentence. In the German equivalent,
however, this has to be in the dative case (with the subject position either filled with the
dummy-subject es (12) or omitted (13)):
(12) Es ist ihm erlaubt, seine eigenen Pläne zu verfolgen. It is him.DAT allowed his own plans to pursue „He is allowed to pursue his own plans.‟
(13) Ihm ist erlaubt, seine eigenen Pläne zu verfolgen. him.DAT is allowed his own plans to pursue „He is allowed to pursue his own plans.‟
In the learner data, none of the overall 15 attempts at this structure (Year A: 5; Year B: 3;
Year C: 7) are correct. In all cases, the learners transfer the English structure as the
following examples demonstrate:
[a2020_02] Ich glaube, *Kinder *sollten (target: Kindern sollte) erlaubt werden, das Internet zu benutzen,... „I think children should be allowed to use the internet,…‟
[c1059_01] Der Artikel fragt , ob *sie (target: ihnen) erlaubt werden *sollen (target: soll), seine Rache zu bekommen. „The article asks whether they should be allowed to get their revenge.‟
213
It is furthermore worth noting that there is a marked overuse of this modal meaning as
a whole in Year C, i.e. modal verbs (DÜRFEN and KÖNNEN) as well as corresponding
PMEs are significantly overused only in this year group. The PME overuse might be
partly influenced by specific topics in the learner data and the way these topics are
phrased. Out of the 54 instances 22 (40%) pertain to the topic of medically assisted
suicide where the exact essay question was: “Sollte aktive Sterbehilfe in Deutschland
gesetzlich erlaubt werden?” (Should medically assisted suicide be legalised in
Germany?). As the lexical modal expression erlaubt already occurs in the question, it is
not surprising that the learners pick up this item and use it more often in their writing on
this specific topic.
This concludes the presentation of periphrastic modal expressions. The analysis of
this large and divergent group of items has brought to light that the learners continually
expand their vocabulary range from one year to the next. Interestingly, in many
categories, epistemic and non-epistemic, a smaller repertoire of items that the learners
are able to use productively (i.e. in Year A) also corresponds to a smaller number of
occurrences of these items overall, whereas an overuse of PME occurrences is typically
found in Year C, where the vocabulary range is broadest. This means that lower
proficiency learners have a more limited range of items and at the same time use these
items less frequently, while higher proficiency learners with a broader range of items also
proportionally use these items much more frequently. One notable exception is the
category that, amongst others, includes the propositional attitude verbs glauben and
denken, where the extreme overuse in Year A gradually decreases almost to native
speaker level in Year C.
Furthermore, it has become clearly evident that the learners tend to use many of the
investigated items repetitively in the same collocational and colligational sequences.
Through the year-by-year comparison it is possible to trace the development of these
items into rigidly set and formulaic phrases that dominate almost all of the meaning
categories in Year C.
5.3 MODAL INFINITIVES
Modal infinitive constructions (HABEN ZU and SEIN ZU) are the smallest group of modal
expressions in both CLEG and KEDS. Table 71 gives an overview of the frequencies:
214
Year A Year B Year C KEDS
raw freq.
per 10,000 words
raw freq. per
10,000 words
raw freq. per
10,000 words
raw freq. per
10,000 words
SEIN ZU 8 1.7 11 1.7 50 5.7 60 7.3
HABEN ZU 0 0.0 8 1.2 1 0.1 14 1.7
TOTAL 8 1.7*** 19 2.9*** 51 5.8* 74 9.0
TABLE 71: OVERALL FREQUENCIES OF MODAL INFINITIVE CONSTRUCTINS IN LEARNER AND NATIVE
SPEAKER CORPORA (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
The frequency figures reveal that all learner groups significantly underuse modal
infinitives compared to the native speaker group.
5.3.1 SEIN ZU
SEIN ZU is the more frequent modal infinitive in both learner and native speaker data.
Both learners and native speakers follow the general observation that SEIN ZU is mostly
used in the meaning of possibility (in competition with the modal verb KÖNNEN). The
native speakers use it in the meaning of necessity (equivalent to MÜSSEN) more often
than any of the learner groups. Other meanings (advice and permission) do occur, but
very infrequently as the following table shows:
Year A Year B Year C KEDS
possibility 6 9 48 37
necessity 1 0 1 10
advice 1 0 1 1
permission 0 2 0 2
ambiguous 0 0 0 10
TOTAL 8 11 50 60
TABLE 72: RAW FIGURES OF SEIN ZU ACROSS SEMANTIC MEANING CATEGORIES
Table 72 reveals that in CLEG, all occurrences of SEIN ZU can unambiguously be
assigned to one of the semantic meaning categories.
[c1056_01] Befürworter der Todesstrafe argumentieren, die Todesstrafe diene dem Heilungsprozess der Angehörigen der Opfer. Aber ist das zu rechtfertigen? ‟Supporters of the death penalty argue that the death penalty serves the healing process of the victim‟s family. But can that be justified?‟
215
[c2044_02] In einem System wo Gebühren später wieder bezahlt müssen spielt der "Sozialstatus" des Students eine geringere Rolle als ein System wo alle Gebühren sofort am Anfang des Studiumjahres zu entrichten sind. ‟In a system where fees have to be paid back later the student‟s “social status” plays a smaller role than in a system where all fees have to be paid at once at the beginning of a study year.‟
Example [c1056_01] unambiguously denotes possibility, while [c2044_02]
unambiguously denotes necessity. In the native speaker data, however, we encounter
examples that display exactly the kind of ambiguity between the different modal
meanings that is uniquely characteristic of modal infinitives:
[keds_024] Die Haltlosigkeit in der Gesellschaft , [...], ist laut Ulrich Beck auf die zunehmend kaputten Familienverhältnisse [...] zurückzuführen. ‟According to Ulrich Beck, the instability in today‟s society can/must be attributed to the increasing number of broken homes.‟
[keds_062] Prof. Dr. Petris Idee, die Kinder in den Demokratisierungsprozess einzubeziehen, ist grundlegend zu unterstützen. ‟Prof. Dr. Petri‟s idea of involving children in the democratisation process must/can/should fundamentally be supported.‟
In example [keds_024] the ambiguity is between possibility and necessity, in [keds_062]
even a three-way ambiguity between possibility, necessity and advice is present. The 10
instances of ambiguity in KEDS are equivalent to 16% of the overall frequency of SEIN
ZU – the same percentage as those instances that can be unambiguously assigned to
the meaning of necessity.
On top of this, two further differences between learners and native speakers can be
observed. First of all, the majority of sentence structures in which the SEIN ZU
constructions are found in the learner data are of the familiar semi-fixed es ist … format.
The two phrases that occur most frequently (usually sentence-initial) are Es ist nicht zu
leugnen, dass… (It cannot be denied that…) and Es ist leicht zu ersehen, dass... (It is
easy to see that…). These phrases are very rare in KEDS:
Year A Year B Year C KEDS
sequence: es ist (ADJ) zu V, dass... 3 6 32 2
as % of all instances of SEIN ZU 37.5 54.4 64.0 3.3
TABLE 73: COLLIGATIONAL SEQUENCES ES IST (ADJ) ZU V, DASS…
The fact that most of the occurrences of SEIN ZU are „tied up‟ in these semi-fixed
phrases also means that the distribution of types of verbs that the construction is used
216
with in learner and native speaker corpora is different. In the learner data, especially in
Year C, the majority of infinitives are verbs that express subjective attitude (e.g. leugnen
- deny, glauben - believe) and verbs of sensory perception (e.g. (er)sehen – see ). In
KEDS, however, the biggest group of verbs is mental/speech act verbs (e.g.
unterscheiden - distinguish; bedenken - consider) and physical action verbs (e.g.
bestrafen - punish; unterbinden - stop/prevent). The following table provides a summary:
Year A Year B Year C KEDS
attitude verbs 3 7 22 3
sensory perception 0 0 13 10
mental/speech act 0 3 8 32
physical action 5 1 6 14
TABLE 74: RAW FREQUENCIES OF INFINITIVE VERB TYPES IN SEIN ZU CONSTRUCTIONS
The second difference between learners and native speakers is that learners overuse
SEIN ZU constructions that are modified further by an adjective. These adjectives
indicate that the action captured in the verb phrase is either easy or difficult to achieve,
i.e. the occurring adjectives are leicht/einfach/klar (easy, clear) and schwer (difficult,
hard) as in the following examples:
[a1019_03] Ein Erstjahr Student erklärt, dass ein Lernprogramm sehr einfach zu benutzen ist. ‟A first-year student explains that an e-learning program is very easy to use.‟
[c2045_06] Es ist schwer zu sagen, ob jemand eine solche Krankheit überleben wird ‟It is hard to say whether someone will survive such an illness.‟
Table 75 gives the frequencies of these modal infinitive constructions that are further
modified by an „easy/difficult‟ adjective:
Year A Year B Year C KEDS
modification by ADJ: einfach 3 1 16 2
modification by ADJ: schwer 0 2 4 1
TOTAL 3 3 20 3
as % of all instances of SEIN ZU 37.5 27.3 40.0 5.0
TABLE 75: RAW FREQUENCIES OF COLLIGATIONAL SEQUENCE SEIN ADJ ZU INF
217
Looking at the English translation we can infer that the high frequency of the German
SEIN ADJ(difficulty) ZU INF sequences is due to transfer from the common English
colligational sequence BE ADJ(difficulty) to INF130.
5.3.2 HABEN ZU
HABEN ZU is used almost as much as SEIN ZU in Year B, but otherwise this
construction does not figure much in the learner data. Although it is also not used often
in KEDS, the learner data still shows an underuse of this construction (cf. Table 71
above). From a contrastive point of view this is surprising insofar as the English
equivalent structure (have to) is much more common than the German counterpart (and
in certain contexts compulsory as the replacement for the incomplete paradigm of the
modal verb MUST, e.g. in past tense). A count in the English argumentative essay
corpus LOCNESS results in 232 instances of have to, which equals 7.2 instances per
10,000 words. This is roughly seven times more than the highest frequency for German
HABEN ZU (1.2 in Year B) in CLEG, but also over four times more than the frequency of
HABEN ZU in KEDS. A possible explanation would be that the learners are aware that
English have to has a much wider application range and therefore avoid the German
construction, as they do not „trust‟ its transferability even in the cases where it would be
possible. The low frequency of this item in the German language should also contribute
to the low salience for the learners. Another plausible explanation could be that in the
teaching material learners are introduced to the modal verb MÜSSEN as the prototypical
translation of English have to, as in German the modal verb is the prototypical means of
expressing necessity whereas in English the use of the modal verb MUST is more
restricted (see 2.3.4).
In terms of ambiguity the same observation as with SEIN ZU can be made, namely
that all instances in CLEG are unambiguous and all bar one are used for the more
common meaning of necessity, rather than possibility, as the following example
demonstrates:
[b1004_04] Frauen leben zunehmend, um zu arbeiten, weil sie vielleicht fühlen, dass sie etwas zu beweisen haben. ‟Women increasingly live for their work, maybe because they feel that they
130
A search of this pattern in the BNC with the adjectives „difficult, hard, easy, clear‟ returns 7981 hits (81.2 per million words).
218
have something to prove.‟
In KEDS, 13 out of 14 instances of HABEN ZU are used in the sense of necessity, only
one example of possibility occurs; there are also no examples for ambiguous HABEN ZU
constructions.
In summary, we have seen that both SEIN ZU and HABEN ZU constructions are
underused by the learners. We have, however, also seen qualitative differences, which
can be partly assigned to an excessive use of semi-fixed es ist … zu phrases and partly
to a suggested transfer of English semi-fixed phrases of the it is easy to see kind.
HABEN ZU constructions, on the other hand, are by and large avoided by the learners
possibly because of concern of negative transfer.
5.4 SUBJUNCTIVE MOOD (KONJUNKTIV)
5.4.1 Overview
Overall, 917 subjunctive forms were detected and annotated in CLEG, 358 in KEDS.
These figures include subjunctives of lexical verbs and the auxiliary verbs haben, sein
and werden131, but exclude subjunctive forms of modal verbs. As subjunctives of modal
verbs (especially Konjunktiv II forms) were already discussed in conjunction with the
modal verbs above (see Chapter 5.1), they will only be discussed peripherally in this
section (e.g. in the section on conditional clauses below). For the sake of completeness,
however, they are included in all further counts of subjunctive forms. Table 76 and Table
77 below give an overview of the distribution of Konjunktiv I and Konjunktiv II forms in
the different subcorpora:
131
Due to the fact that werden can function as main verb, auxiliary verb and a semi-modal verb as well as providing the analytical subjunctive form (würde + bare infinitive), this verb is counted together with lexical verbs and AUX verbs rather than modal verbs.
219
Year A Year B Year C KEDS
raw freq.
per 10,000 words
raw freq.
per 10,000 words
raw freq.
per 10,000 words
raw freq.
per 10,000 words
KON I lexical/AUX verbs
20 4.3** 81 12.3* 77 8.8 69 7.5
KON I modal verbs
4 0.9 16 2.4* 18 2.1* 8 1.0
TOTAL 24 5.2** 97 14.7** 95 10.8 77 8.5
TABLE 76: OVERALL FREQUENCIES OF KONJUNKTIV I FORMS IN LEARNER AND NATIVE SPEAKER
CORPORA (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
Year A Year B Year C KEDS
raw freq.
per 10,000 words
raw freq.
per 10,000 words
raw freq.
per 10,000 words
raw freq.
per 10,000 words
KON II lexical/AUX verbs
112 24.3*** 173 26.3** 471 53.8*** 290 31.5
KON II modal verbs
127 27.5 128 19.4*** 517 59.0*** 266 32.4
TOTAL 239 51.8*** 301 45.7*** 989 112.8*** 556 63.9
TABLE 77: OVERALL FREQUENCIES OF KONJUNKTIV II FORMS IN LEARNER AND NATIVE SPEAKER
CORPORA (* = SIGNIFICANT AT P<0.05; ** = SIGNIFICANT AT P<0.01; *** = SIGNIFICANT AT P<0.001)
We can see that Konjunktiv I with lexical or auxiliary verb is underused significantly in
Year A compared to the native speaker data, but overused in Year B. Year C sees a
highly significant overuse of Konjunktiv II forms, whilst there is significant underuse in
Years A and B. A detailed look into the different categories should reveal a clearer
picture of, and give clues to, possible reasons for these patterns. Before going into these
details, however, it is necessary to turn our attention to some errors with the Konjunktiv.
Short of manually looking at every individual verb form in its context (ca. 24,000 in
CLEG), it is impossible to determine errors where an indicative verb form is used in a
context where a Konjunktiv form is compulsory. Also, as the Konjunktiv is only
compulsory in a very limited set of circumstances (mainly retrospective counterfactual
conditionals), in most circumstances it is a matter of stylistic choice whether to use a
Konjunktiv or not (cf. 2.3.5.2). Owing to the much smaller numbers and the fact that
Konjunktiv forms were manually annotated, it is, however, possible to determine those
220
errors where a Konjunktiv form is used instead of an indicative form. The following table
gives a count of this error type (including errors on modal verbs):
Year A Year B Year C
raw freq. % of all KON I/II
raw freq. % of all KON I/II
raw freq. % of all KON I/II
KON I instead of indicative 14 58.3 27 27.8 28 29.5
KON II instead of indicative 21 8.8 51 16.9 42 4.3
TABLE 78: ERROR FREQUENCIES: KONJUNKTIV INSTEAD OF INDICATIVE VERB FORMS
As the table shows, the error rate is highest for Konjunktiv I forms in Year A (58.3%). Out
of the 14 errors here, 8 are down to confusions involving werde (Konjunktiv I of
WERDEN). This is used, in equal proportions, instead of the preterite indicative form
wurde or the indicative modality/future tense marker wird, as in the following examples:
[a2025_04] Aber vor den zweiten halb der zwanziger Jahrhundert Deutschland *werde (target: wurde) sehr Nationalistisch und es hat viele undemokratische Regierungen. „But before the second half of the 20th century German became very nationalistic and it has many undemocratic governments.‟
[a1015_02] Diese Deutsche glauben, dass ihre Sprache nicht überschattet sein sollte. Aber *werde (target: wird) es schwer sein, die Benutzung von Englisch unter Jugendlichen an zu halten, weil es als "kool" schätzt ist. „These Germans believe that their language should not be overshadowed. But it will be difficult to stop young people using English, because it is considered “cool”‟.
This type of error also makes up the biggest group of errors in Konjunktiv I in Years B
and C. Taking into account that a considerable proportion of Konjunktiv I occurrences
are erroneous, this means that genuine Konjunktiv I forms in Year A are even more
significantly underused132, while the overuse in Year B becomes non-significant133.
As the main domain of the Konjunktiv I is indicating indirectness (e.g. in indirect
speech) rather than modality, and since this is the only function in which the Konjunktiv I
was found to be used in the learner data, it will not be pursued any further here. Instead
the focus of the investigation will be on the Konjunktiv II.
Errors with the Konjunktiv II, where an indicative form would be appropriate, do not
feature as frequently in the learner corpora. Surprisingly, the highest error rate with
132
LL-value: 27.06 (significant at p<0.001); LL-value before deducting errors: 6.99 (significant at p<0.01). 133
LL-value: 0.57 (not significant); LL-value before deducting errors: 8.81 (significant at p<0.01).
221
Konjunktiv II occurs in Year B. 50% of these errors are uses of the Konjunktiv II of the
modal verb KÖNNTE, instead of either the indicative present tense form KÖNNEN or the
indicative preterite form KONNTE, e.g.:
[b1043_02] Die meisten Studenten, die ich interviewiert haben, hatten ziemlich ähnlich vorgefasste Meinungen über den Lebensstil der Engländer bevor sie nach Lancaster gekommen sind, damit sie hier studieren und wohnen *könnten (target: können). „Most of the students who I interviewed hat similar preconceptions about the way English people live before they came to Lancaster so they could live an study here.‟
[b1031_01] In jeder Besatungzone war die Regierung verantwortlich für diese Zeitungen. Aus diesem Grund *könnten (target: konnten) sie veröffentlichen, was sie wollten. „In every occupied zone the government was responsible for these newspapers. For this reason they could publish whatever they wanted.‟
While the example above is quite clearly a case of morphological confusion (not
helped by the fact that the English form COULD is ambivalent between past and
subjunctive form), in many cases it is hard to determine whether these errors are due to
morphological confusion or due to semantic problems, i.e. determining when a
subjunctive form is semantically appropriate, as in the following example:
[c2032_06] Manche Länder erlauben schon die Sterbehilfe. Manche noch nicht . Das heißt, Sterbende in der Schweiz oder Belgien *könnte (target: können) die Sterbehilfe bekommen aber nicht die Sterbenden in anderen Ländern wie England. „Some countries allow medically assisted suicide. Some do not yet. This means that dying patients in Switzerland or Belgium *could (target: can) receive euthanasia but not the dying patients in other countries like England.‟
As with the Konjunktiv I, another common error type with Konjunktiv II involves the
morphology of WERDEN, where learners confuse wurde and würde, e.g.:
[a1014_03] Die deutschen Ausbildungsbehörden verlangen nach eine Kampagne , dass Deutsch als ein Shulfach fördern *wurde (target: würde). „The German education authorities call for a campaign that would promote German in schools.‟
[c2029_05] Aktive Sterbehilfe ist wie eine Einstellung zum Leben gesehen, und wenn viele Patienten darüber sprechen , haben sehr viele diese Menschen Angst , dass sie getötet werden *wurden (target: würden). „Active euthanasia is seen as an attitude to life and when many patients talk about it they are afraid that they might be put to death.‟
222
5.4.2 Lexical distribution of Konjunktiv II
Table 79 displays the distribution of Konjunktiv II forms across lexical verbs, auxiliary
verbs134 (hätte, wäre), würde and modal verbs.
Year A Year B Year C KEDS
raw freq.
per 10,000 words
raw freq.
per 10,000 words
raw freq.
per 10,000 words
raw freq.
per 10,000 words
modal verbs 127 27.5 128 19.4 517 59.0 266 32.4
würde 71 15.4 105 15.9 144 16.4 143 17.4
AUX verbs 38 8.2 58 8.8 297 33.9 121 14.7
lexical verbs 3 0.7 10 1.5 30 3.4 26 3.2
TOTAL 239 51.8 301 45.7 988 112.8 556 67.7
TABLE 79: DISTRIBUTION OF KONJUNKTIV II FORMS ACROSS VERB CATEGORIES
We can see that in all learner groups as well as in the native speaker group lexical verbs,
as expected, are least common in the Konjunktiv II, with modal verbs being the most
common group. A look at the percentages that each of these types of verbs make up out
of the overall Konjunktiv II frequencies (Table 80 below) reveals that the native speakers
use the biggest proportion of lexical verbs in the Konjunktiv II compared to the learner
groups. The proportion of würde-Konjunktiv is highest in Years A and B; in Year C it is
considerably lower than in KEDS, with bigger proportions taken up by AUX and modal
verbs.
Year A Year B Year C KEDS
modal verbs 53.1% 42.5% 52.3% 47.8%
würde 29.7% 34.9% 14.6% 25.7%
AUX verbs 15.9% 19.3% 30.1% 21.8%
lexical verbs 1.3% 3.3% 3.0% 4.7%
TOTAL 100% 100% 100% 100%
TABLE 80: KONJUNKTIV II VERB CATEGORIES AS PERCENTAGES OF OVERALL KONJUNKTIV II FREQUENCIES
The proportions of Konjunktiv II in the different verb categories in KEDS conforms
largely with an older corpus investigation by Jäger (1971), who used the Mannheimer
134
The label “auxiliary verb” is used here for all uses of hätte and wäre, regardless of whether they are actually used as auxiliary verbs or main verbs.
223
Korpus135 to determine that, across all text types in that corpus, 8% of all Konjunktiv II
forms were lexical verbs, 25% würde and 66% AUX verbs and modal verbs. The slightly
lower figure for lexical verbs in KEDS can either be attributed to a general decline in
synthetic Konjunktivs in contemporary German (cf. Fabricius-Hansen 1997:17) or to the
writing style of pupils, which might not be considered as accomplished as some of the
professionally written texts in the Mannheimer Korpus. In the learner data, it is evident
that there is a progression from Year A to Year B in terms of lexical verb Konjunktiv II
forms, although proportionally it stagnates at a level that is below the native speaker
group. The following lexical verbs in the Konjunktiv II occur in the individual sub-corpora:
Year A Year B Year C KEDS
gäbe 2 gäbe 1 anschaute 1 anböte 1
*gabe 1 bekäme 2 benähme 1
käme 1 erreichte 1 bestünde 1
kotzte 1 gäbe 15 bliebe 1
*taugten 1 ginge 2 ergäbe 1
*tränkten 1 handelte 1 gäbe 4
träumte 1 hieße 1 ginge 1
*verständen 1 klonte 1 hieße 1
wüsste 2 läge 1 käme 4
litte 1 läge 2
sähe 1 ließe 4
stünde 1 redete 1
verstände 1 schaffte 1
wüsste 1 stünde 1
unterschiede 1
wüsste 1
TOTAL 2 TOTAL 10 TOTAL 30 TOTAL 26
TABLE 81: KONJUNKTIV II LEXICAL VERB TYPES AND RAW FREQUECY FIGURES (* = ERRORS; VERBS
THAT OCCUR IN CLEG AND KEDS IN BOLD)
This table indicates that there is a gradual increase in the range of lexical verbs that are
used in their Konjunktiv II form from Year A to Year C. This progression can be summed
up as follows: In Year A, Konjunktiv II with lexical verbs is largely avoided; in Year B,
more lexical verbs in Konjunktiv II occur, but with that comes also a considerable amount
of errors (40%); in Year C, the learners are using a range of verbs close to that of the
native speakers and manage to avoid errors completely. We now turn our attention from
135
Ca. 2.2 Million words from texts originating 1950-1967. Text types: fiction, scientific and popular scientific articles, light fiction, newspaper and magazine articles.
224
the lexical choices of Konjunktiv II forms to the types of contexts in which the Konjunktiv
II is used.
5.4.3 Context types of Konjunktiv II
The following overview chart gives a summary of the different types of modality contexts
that the Konjunktiv II is used for in the learner and native speaker data:
FIGURE 19: FREQUENCIES OF KONJUNKTIV II CONTEXT TYPES
Figure 19 shows that the distribution of Konjunktiv II context types is similar for learners
and native speakers, i.e. Konjunktiv II occurs most frequently in sentences that are
involved in non-factual argumentation, then in conditional clauses, then in indirect
speech contexts (cf. 2.3.5.3). Modality contexts where Konjunktiv II is used as a marker
of politeness are rare. Other Konjunktiv II contexts, such as counterfactual comparative
clauses, purpose clauses, counterfactual relative clauses and counterfactual consecutive
clauses only occur very sporadically both in the leaner and native speaker data. We can
see, however, that there is an underuse of Konjunktiv II non-factual argumentation
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
freq. per
10,000 words
Year A
Year B
Year C
KEDS
non-factual
argumentation
conditionalsindirect speechpolitenessothers errors (indicative
required)
225
contexts in Years A and B compared to KEDS, while this type is overused in Year C (all
three are statistically highly significant at p<0.001). Conditionals with Konjunktiv II are
overused significantly in Year C; the overuse in Years A and B are not statistically
significant compared to KEDS.
If we look, however, at the percentage distribution of types of overall Konjunktiv II
frequencies (Figure 20), it becomes clear, that in the learner corpora roughly twice as
many of all Konjunktiv II forms occur in conditional clauses than in KEDS (Year A:
20.5%; Year B: 20.9%; Year C: 26.6%; KEDS: 11.3%). At the same time, the proportion
of non-factual argumentation is considerably smaller for all learner subcorpora compared
to KEDS (Year A: 56.1%; Year B: 45.5%; Year C: 62.1%; KEDS: 77.5%):
FIGURE 20: DISTRIBUTION OF KONJUNKTIV II ACROSS CONTEXT TYPES AS PERCENTAGES OF OVERALL
KONJUNKTIV II FREQUENCIES
In order to get to the bottom of this discrepancy of Konjunktiv II uses in conditional
clauses, we shall look at these in more detail.
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
% of all KII uses
Year A
Year B
Year C
KEDS
non-factual
argumentation
conditionalsindirect speechpolitenessothers errors
(indicative
required)
226
5.4.4 Konjunktiv II in conditional clauses
Overall, 944 sentences containing conditionals were detected and analysed in CLEG,
281 in KEDS. These distribute over the different subcorpora as follows:
Year A Year B Year C KEDS
conditionals (raw frequencies) 203 226 515 281
overall number of sentences 2839 3792 4814 4249
conditionals as % of overall sentences
7.2 6.0 10.7 6.6
TABLE 82: OVERALL FREQUENCIES OF CONDITIONALS
Table 82 shows that, while the proportion of conditional clauses is similar to KEDS for
Years A and B, there is a marked overuse of conditional clauses overall in Year C. We
know from Figure 20 that Konjunktiv II is overused in conditional clauses in all year
groups, so if we look at the proportional distribution of different types of conditional
clause patterns (Figure 21) it is not surprising that the ones containing Konjunktiv II are
overused by the learners:
FIGURE 21: CONDITIONAL CLAUSE PATTERNS AS % OF OVERALL FREQUENCIES OF CONDITIONALS
The categories are encoded as follows: P and A refer to protasis (if-clause) and apodosis
(main clause containing the consequence of P) respectively. (I) indicates indicative mood
(any tense) in the designated clause, (KII) indicates Konjunktiv II in that clause. The
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
% of all
conditionals
Year A
Year B
Year C
KEDS
P(I)-A(I)P(KII)-A(KII)P(I)-A(KII)P(KII)-A(I)other
227
labelling is regardless of which clause occurs first in the actual sentence. The four main
categories are therefore:
P(I) – A(I):
[a2016_03] Das Internet ist nützlich , wenn es in der richtigen Art benutzt wird aber es soll das einzige Hobby für Kinder nicht sein. „The internet is useful if it is used in the right way but it should not be the only hobby for children.‟
P(KII) – A(KII):
[c1058_02] Wenn Benimm-Kursen eingefüht würden, gäbe es nicht so viel Zeit auf den Stundenplan für akademische Klassen wie Fremdsprachen oder Wissenschaft. „If good manners courses were introduced, there wouldn’t be much space in the timetable for academic classes like foreign languages or science.‟
P(I) – A(KII):
[keds_011] Auch wenn der Staat auf eine spezielle Besteuerung in der Art der Tabak- und der Alkoholsteuern verzichtet, würde der Handel in jedem Fall von der Umsatzsteuer erfaßt. „Even if the government refrains from imposing a special kind of tax like the one on tobacco or alcohol, the trade would in any case be liable to VAT.‟
P(KII) – A(I):
[c2044_01] Wenn eine Krankheit früher diagnostiziert werden könnte, gibt es für den Patient weniger Belastung. „If an illness could be diagnosed earlier, there is less strain for the patient.‟
Figure 21 shows that by far the majority of conditionals in all subcorpora is of the type
P(I) – A(I)136, containing indicatives in both main and if-clause (or, in Nehls‟ (1978)
terminology, „direct prospective predicative‟ conditionals, see 2.3.5.3). These are not our
main concern here, as our interest lies mainly with the types containing Konjunktiv II. It
is, however, noteworthy that there are some signs of L1-influence in this type of
conditional. In English, the default sequence of tenses is P(present tense) – A(will), e.g.:
136
Year A: 77.8%; Year B: 74.3%; Year C: 62.1%; KEDS: 84.0%
228
(14) If you park your care here, you will get a ticket.
In German, on the other hand, the sequence is P(present tense) – A(present tense or
WERDEN), with the present tense in the apodosis being the much more common option
than WERDEN, e.g.:
(15) Wenn du dein Auto hier parkst, bekommst du einen Strafzettel.
In KEDS, the proportion of WERDEN in the apodosis out of all direct prospective
predicative conditionals is 5.5%. A look at the learner data reveals that learners do
overuse WERDEN in apodosis clauses compared to KEDS, thus following the default
English pattern rather than the German one. This overuse, however, is not so great as to
warrant assigning the overuse to the transfer of this conditional clause pattern: Year A:
8.2%; Year B: 8.9%; Year C: 7.5%. Considering that a general L1-influenced overuse of
the verb WERDEN was already established in the modal verb section (cf. 5.1.1.6), we
are probably seeing the effects of this here rather than an overuse induced specifically
by this particular conditional clause pattern.
Going back to the overuse of Konjunktiv II in conditionals, Figure 21 shows a
considerable overuse of two patterns containing Konjunktiv II: P(KII) – A(KII)137 and
P(I) – A(KII)138. The conditional pattern containing only indicatives (P(I) – A(I)) is slightly
underused by the learners compared to the native speaker group; this seems to suggest
that the learners have more of a tendency to explicitly mark modality/non-factuality in
conditional clauses by using Konjunktiv II forms than the native speakers (although, as
discussed in 2.3.5.3.a, indicative in conditional clauses does not necessarily indicate
factuality). We will now look into these patterns that contain Konjunktiv II.
In German, the pattern P(KII) – A(KII) is used for indirect prospective predicative
conditionals (16) or non-factual prospective conditionals (17):
137
Year A: 13.8%; Year B: 15.0%; Year C: 19.2%; KEDS: 5.3% 138
Year A: 7.4%; Year B: 9.3%; Year C: 16.3%; KEDS: 7.1%
229
(16) Wenn er seine Hausaufgaben gewissenhaft machen würde, würde er bessere Noten bekommen. „If he did his homework more thoroughly, he would get better marks.‟
(17) Wenn morgen Wahlen wären, würde die SPD verlieren. „If elections were held tomorrow, the SPD would lose.‟
The second pattern, P(I) – A(KII) can be said to be a mixture between direct and indirect
prospective predicative conditionals (18):
(18) Wenn er ein Auto stiehlt, würde er ins Gefängnis gehen. „If he stole a car, he would go to prison.‟
If we look at the translations of the last three examples we can see that the English
pattern for all of these is a variation of the German P(I) – A(KII) pattern139.
The difference from the German pattern is, however, that there is a tense-shift to past
tense140 in English, whereas in the German pattern with indicative, i.e. the mixed
direct/indirect pattern (18), there is no tense-shift in the protasis, i.e. the wenn-clause
contains an indicative present tense form (a past tense form would indicate posteriority,
which is neither intended in example (18) nor in the following corpus example illustrating
the mixed direct/indirect pattern):
[keds_046] Außerdem läge ein Verstoß gegen den Grundsatz der Verhältnismäßigkeit vor, wenn Hunde der K1 trotz bestandenem Wesenstest zum Tragen eines Maulkorbs verpflichtet sind. „It would furthermore amount to a violation of the principle of proportionality if dogs of the category K1 were obligated to wear a muzzle even after passing a character test.‟
Secondly, this pattern cannot be used for non-factual prospective conditionals, as for
these a Konjunktiv II form is required in both protasis and apodosis, e.g:
[keds_044] Was würde geschehen, wenn es plötzlich keine Noten mehr geben würde? „What would happen, if marks were suddenly not used anymore?‟
Looking at the error rates for these two types in the learner data, we can see that both of
these patterns cause considerable problems especially for the less advanced learners:
139
A translation involving the English subjunctive in the protasis is conceivable in these contexts („If he were to do his homework more thoroughly…‟ – „If elections were to be held tomorrow…‟ – „If he were to steal a car…‟). The use of the subjunctive in conditional clauses is, however, very rare in present day English (cf. Erdmann 1981:115) and is mainly restricted to historically fossilised formulaic phrases, e.g „If I were you…‟.
140 According to Nehls (1978:52) this tense-shift acts as a tense-metaphor (where “remote” tense, i.e. past tense, equates to “remote” factuality, i.e. non-factuality) within the otherwise present tense context.
230
Year A Year B Year C
frequency pattern P(I) – A(KII) 15 21 84
errors 5 7 17
errors as % of all P(I) – A(KII) 33.3 33.3 20.2
TABLE 83: ERROR RATE FOR CONDITIONAL CLAUSE PATTERN P(I) – A(KII)
Year A Year B Year C
frequency pattern P(KII) – A(KII) 28 34 99
errors 15 16 19
errors as % of all P(KII) – A(KII) 53.6 47.1 19.2
TABLE 84: ERROR RATE FOR CONDITIONAL CLAUSE PATTERN P(KII) – A(KII)
From the cross-linguistic differences in the conditional patterns that have just been
discussed, one might expect errors mainly in the form of indicative past tense forms in
the protasis (if-clause) instead of either indicative present tense or Konjunktiv II. A closer
look at the individual errors, however, reveals a different reality.
All errors in the P(I) – A(KII) patterns in all three learner groups occur in the apodosis
clause, i.e. on the Konjunktiv II form. These errors consist almost exclusively of errors on
modal verbs, specifically KONNTE instead of KÖNNTE , SOLLEN instead of SOLLTE
and the past tense form of WERDEN (wurde) instead of würde (Year A: 100%; Year B:
100%; Year C: 88%). We have already seen that the distinction between the preterite
and subjunctive forms of modal verbs (cf. Table 18, p.135), the semantic distinction
between SOLLEN and SOLLTE (cf. Table 27, p.154) and the morphology of WERDEN
(cf. 5.1.1.6 and 5.4.1) are common problems for the learners. This suggests that the
problem here could be, at least in parts, not specifically to do with the conditional clauses
themselves, but rather a more general difficulty with various morphological verb forms
and their semantic differences. This observation runs contrary to Kufner‟s (1962:77)
statement that “most of the difficulties which our English-speaking students encounter
are of a semantic nature, i.e. they have more trouble determining when and why to use a
subjunctive than how to use it.”
The picture is slightly different in the second pattern, P(KII) – A(KII). Modal verbs do
not feature frequently in the error counts here. The wurde/würde distinction is only a
major problem in Year A (38% of errors). Instead the errors are of a much more varied
231
kind: learners, especially in Years A and B, use indicative forms – both present and
preterite – of auxiliary verbs, modal verbs and main verbs. The following examples serve
to show this diversity:
[a2020_02] Wenn es keine Kontrolle des Internets *gab (target: gäbe), dann würden Eltern viele Probleme haben, ... „If control over the internet didn‟t exist, parents would have many problems…‟
[a2018_04] Meinung nach wäre es besser wenn alle Leute besser integriert *sind (target: wären). „In my opinion it would be better if all people were better integrated.‟
[b1006_05] Wenn wir die Lotterie gewinnen würden, könnten wir mit keiner Arbeit überleben, oder wenn wir eine reiche Familie hätten, *brauchen (target: bräuchten) wir kein mehr Geld. „If we won the lottery, we could survive without work, or if we had a rich family, we wouldn‟t need any more money.‟
[c2033_02] Wenn wir keine Demokratie *haben (target: hätten), würden wir die meisten unserer Sozialrecht verlieren. „If we didn‟t have a democracy, we would lose most of our social rights.‟
Another point worth noting is that in Year C, a third of the errors (6 out of 19; 32%) are
actually Konjunktiv II forms where indicative forms are required. This type of error does
not occur in Years A and B.
[c2045_01] Es bleibt zu hoffen, dass die Gentechnik nicht außer Kontrolle entgleiten *würde (target: wird), wenn sie nicht gestoppt werden *würde (target: wird). „It remains to be hoped that genetic engineering won‟t get out of control if it isn‟t stopped.‟
[c1051_05] Trotzdem ist es deutlich, das es nicht möglich wäre, Kosten für Studenten zu reduzieren, und gleichzeitig Facilitäten zu verbessern. Wenn man Universitätqualifikationen kriegen will, *müsste (target: muss) man leider bezahlen. „It is nevertheless obvious that it wouldn‟t be possible to reduce costs for students and improve facilities at the same time. If you want to get university qualifications, unfortunately you have to pay.‟
In this conditional pattern, more errors occur in the protasis (if-clause) clause than in the
apodosis clause, as the following table demonstrates:
232
Year A Year B Year C
errors in protasis clause 16 12 14
errors in apodosis clause 6 6 10
TABLE 85: ERRORS IN CONDITIONAL CLAUSE PATTERN P(KII) – A(KII)
Together with the types of errors that occur in the P(KII) – A (KII) pattern the figures from
Table 85 might be taken as evidence that learners do have difficulties with the fact that,
in German non-factual conditional clauses, both the protasis and the apodosis clause
require a Konjunktiv II form. This is most probably in addition to grappling with the
morphology of the Konjunktiv II itself. In this case, the corpus data cannot provide further
clues as to which of the underlying causes for problems is more prevalent (morphology
or semantics). Other means of gathering information on this issue, e.g. elicitation or
think-aloud tests, might yield more definitive information.
In the last section, we have mainly concentrated on errors rather than on positive
performance, as this has proven to give some indication of the difficulties involved in the
acquisition and correct application of the Konjunktiv. It should have become clear that
these arise due to the very diverse morphological structure of the German Konjunktiv
system as well as some grammatical/semantic aspects of when to use a Konjunktiv –
both aspects where English differs radically from German. This should not, however,
deflect from the fact that learners – even in Year A – have already started to acquire and
understand the German subjunctive system and in many cases achieve correct and
appropriate uses of the Konjunktiv in hypothetical argumentation, conditionals, indirect
speech and other, more marginalised, contexts. To conclude this chapter, we shall
therefore display some examples of this:
[a2008_02] Spätere Öffnung Zeiten würde bedeuten, dass Leute, die spät (wie zum Beispiel Krankenschwestern, Doktoren, Rausschmeißer) arbeiten, ein Getränk nach der Arbeit mit ihren Freunden genießen könnten. „Extended opening hours would mean that people who work late (such as, for instance, nurses, doctors, bouncers) could enjoy a drink after work with their friends.‟
[c2039_05] Niemand hat das Recht Gott zu spielen. Die Sterbehilfe wäre gegen alle Religionen, die behaupten, dass nur Gott das Recht das Leben und das Tod zu kontrollieren hat. „Nobody has the right to play God. Assisted suicide would go against all religions that claim that only God has the right to decide over life and death.‟
233
[a2009_03] Zunächst ist Tony Blair fast daran überzeugt, dass die verlängerten Öffnungszeiten besser für England wäre. „First of all, Tony Blair is absolutely convinced that extended opening hours would be better for England.‟
[b1037_01] Tatsächlich der Artikel sagt, in den USA hätten 60% der Karrierefrauen ihre Jobs verlässen, und in Großbritannien könnte es höher sein. „In fact the article says that in the US 60% of career women had left their jobs and in Great Britain it could be more.‟
[c1056_06] Es scheint als ob, die Möglichkeit nach dem Studium zu zahlen die beste Lösung wäre. „It seems as if the possibility of paying after your course of studies would be the best solution.‟
[b1032_02] Zum Abschluss würde ich sagen , dass das Leben der modernen Frau schieriger geworden ist , aber Frauen haben für Gleichberechtigung gekämpft, und es ist was sie wollen. „In conclusion I would say that the life of modern women has become more difficult, but women have fought for equality and it is what they want.‟
This last part completes the analysis of modality in my learner corpus. The following
chapter is dedicated to evaluating the observations that have been made about various
groups of modal expressions within the context of previous research on modality and
current second language acquisition theories.
234
6. Summary and Discussion
In the previous chapter, the results of the explorative learner corpus investigation were
presented according to the four groups of linguistic means that had been identified as
expressions of modality in the theoretical outline. This chapter consists of, first of all, a
summary of the key results, where links between the different modal means are also
established and, secondly, a discussion of these key findings in light of existing research
on learner modality and second language acquisition theories. One of the main points
that arises out of this discussion is that the linguistic and acquisitional theories that have
to be considered in order to explain the different patterns and phenomena of learner
modality are as diverse as the modal expressions themselves. Cross-linguistic influences
will be shown to be an important explanatory factor as well as intralingual developmental
factors and teaching materials.
6.1 SUMMARY OF RESULTS
As in the results chapter, there are two overarching themes in this summary that
determine the focus of all the different findings. These pertain to results that characterise
similarities and differences between learners and native speakers on the one hand and
between learners at different stages of proficiency on the other. As these two aspects
are often interlinked, in most cases no single factor but an accumulative effect of several
factors provides the key for explaining certain learner specific features. These features
manifest themselves in the following key findings:
1. General overuse of modal expressions:
Learners in all year groups considerably overuse modal verbs. Periphrastic modal
expressions are slightly underused by learners in the lower proficiency groups, but
overused in the advanced group. If all the different modal expressions are
considered together, learners exhibit a tendency to overuse modal expressions, i.e.
they tend to explicitly express modality more than native speakers. This applies in
particular to the most advanced learners, where the propensity to be explicit is also
visible in the overuse of the Konjunktiv II.
235
2. Ranking of modal verbs:
The frequency ranking of modal verbs for all learner groups together is virtually the
same as in the comparable native speaker corpus (KEDS) and the general native
speaker corpus (LIMAS), but split into year groups there are differences:
Lower proficiency groups (Years A and B):
KÖNNEN – MÜSSEN – WOLLEN – SOLLEN – MÖGEN – DÜRFEN
Advanced proficiency group (Year C):
KÖNNEN – SOLLEN – MÜSSEN – WOLLEN – DÜRFEN – MÖGEN
KEDS: KÖNNEN – MÜSSEN – SOLLEN – WOLLEN – DÜRFEN – MÖGEN
3. Form/meaning equivalents in the learners‟ native and target language:
Epistemic modal verbs with form/meaning equivalents in the target language and
high mutual translatability (MÜSSEN, KÖNNEN/KÖNNTE, WERDEN) have been
acquired by learners in all year groups and tend to be overused.
Modal verbs without form/meaning equivalents (evidential SOLLEN and WOLLEN,
MÜSSTE, DÜRFTE) or low translatability (MÖGEN and SOLLTE) have not been
acquired even in the most advanced year group. Only SOLLTE shows some signs
of being acquired in this group.
There is no correspondence between form/meaning equivalents and over-
/underuse patterns in non-epistemic modality. Non-equivalent modal verbs are not
avoided; indeed, most non-epistemic modal verbs are overused by the learners,
but correspondence patterns between modal verbs and periphrastic modal
expressions are similar to those for epistemic modality (see next point).
4. Correspondences between usage patterns of modal verbs and periphrastic
modal expressions:
Periphrastic modal expressions are overused to compensate for the semantic gap
left by the not yet acquired epistemic modal verbs MÜSSTE, SOLLTE, DÜRFTE
and MÖGEN in order to express mitigated certainty or possibility. Generally, both
learners and native speakers prefer to use periphrastic modal expressions for
epistemic modality over modal verbs.
Although there is no corresponding pattern of modal verb underuse, analogous to
the relationship between epistemic modal verbs and PMEs, learners tend to
underuse non-epistemic PMEs where the corresponding modal verb has an L1
form/meaning equivalent (MÜSSEN, SOLLEN/SOLLTE, KÖNNEN/KÖNNTE) and
overuse PMEs where a form/meaning equivalent does not exist for the
236
corresponding modal verb (WOLLEN, MÖCHTE, MÖGEN, DÜRFEN). This is
irrespective of whether the corresponding modal verb is overused or underused.
Generally, both learners and native speakers prefer to use modal verbs for non-
epistemic modality over PMEs.
5. Difficulties with morphological and semantic distinction of indicative and
Konjunktiv II modal verb forms:
Learners in all year groups exhibit grave difficulties with the Konjunktiv II forms of
modal verbs (MÜSSTE, SOLLTE, KÖNNTE, DÜRFTE). These difficulties pertain
both to the morphological forms of these verbs and their semantic distinction from
the indicative verb forms.
6. Personalised expressions:
Personalised expressions explicitly indicate writer involvement in an assessment of
propositional validity. Learners show a distinct tendency for overusing personalised
expressions compared to native speakers. This preference is strongest in the lower
year groups and diminishes with growing proficiency.
7. Range of modal expressions and formulaic sequences:
In the use of periphrastic modal expressions, learners continuously expand their
vocabulary range from lower to more advanced proficiency groups, although even
the most advanced of learners only reach a lexical range that is lower than that of
the native speakers (ca. 75% of different types compared to the native speakers).
The growing vocabulary range also corresponds with a higher frequency of PMEs
overall. The same developmental pattern can be observed with the Konjunktiv II.
Learners develop distinct formulaic sequences in the course of their progression.
In the two lower year groups they often appear to be constructed analytically from
lexical items that the learners have encountered. In the most advanced year group
there is overwhelming evidence of fossilised formulaic sequences, which are
greatly overused by the learners compared to the native speakers. These differ
from the formulaic sequences of native speakers in a number of ways.
The rare occurrences of modal infinitives are also restricted to formulaic
sequences, which do not seem to be analysed for their grammatical content. This
means that, although some learners do use these structures, they do not seem to
have any explicit meta-linguistic knowledge of the category „modal infinitives‟.
8. Konjunktiv II as indicator of learner proficiency:
Usage patterns of the Konjunktiv II can serve as indicators of learner proficiency as
237
its use is limited both in terms of frequency and lexical diversity in the first two
years and only expands to its full use in the final year. However, even in that final
year, learners exhibit a marked tendency to apply the Konjunktiv II in „set/learned‟
contexts, predominantly conditional clauses.
6.2 DISCUSSION
6.2.1 Overuse of modal expressions
Learners in all year groups in CLEG significantly overuse modal verbs compared to
native speakers in the comparable corpus KEDS. But this native speaker group also
overuses modal verbs compared to the general native speaker corpus LIMAS. On the
one hand this can be ascribed to the central role of modality in argumentative writing,
which other studies on modality have also pointed out (cf. e.g. Ehlich and Rehbein 1972;
Redder 2001; Aijmer 2002) and which underlines how text-type sensitive and dependent
modality is (cf. also Dagneaux 1995).
On the other hand, a more learner-centred explanation is possible. Hunston (2002)
conducted a learner corpus study on the spoken language of Norwegian teenagers‟ L2
English and also found that the learners overuse modal verbs compared to the spoken
discourse of teenage native speakers. Aijmer (2002) found in her study on learner
English that advanced learners with various L1s (Swedish, German, French) all overuse
modal verbs in their argumentative writing. The same holds for L1-Cantonese learners of
English (Hyland and Milton 1997). This indicates that the overuse of modal verbs is a
general feature of learner language, which is not dependent on the learners‟ L1. Aijmer
(2002:73) points out that high frequencies of certain modal verbs are characteristic for
native speaker spoken discourse141 (cf. also Biber 2006) and that the tendency for
learners of English to include speech-like features in their written work has been attested
in a number of studies on various aspects of language (e.g. DeCock 2000; Meunier
2000; Lorenz 1999; Granger and Rayson 1998; Altenberg and Tapper 1998). One of
these aspects is highly recurrent word combinations (HRWC). DeCock (2000:58) is able
141
A frequency comparison for modal verbs in the spoken and written German resulted in a similar observation for German. In the DWDS corpus (Deutsches Wörterbuch der deutschen Sprache des 20. Jahrhunderts) all modal verbs occur in significantly higher frequencies in spoken language than in written language. (For a description of the DWDS corpora see e.g. Geyken 2009).
238
to show that the difference between spontaneous speech and formal writing is less
marked in learner corpora than in native speaker corpora with regard to the use of
HRWCs, such as it is true that or I would say that, for example. This is manifest both in
the unusually high frequency of recurrent sequences in learner writing and in the fact that
specific usage patterns for longer HRWCs are similar for native speaker speech and
learner speech AND writing, but distinctly different for native speaker writing.
It has to be noted here, that the distinction between „spoken‟ versus „written‟ language
is not as straightforward as the simple reference to the medium of production suggests
(cf. e.g. Wilson 2009). It has been shown that a number of situational factors, such as
level of formality and overall purpose of the text (e.g. to inform, entertain or convince) are
more accurate for predicting the occurrence of certain linguistic features than the
distinction between speech and writing (cf. e.g. Biber 1988; Nakamura 1993; Takahashi
2006). The current discussion is centred on the specific discourse type of argumentative
writing within the instructional setting of a school or university. This requires a relatively
formal presentation of arguments that prove to the teacher that a student can formulate
their opinion in a refined and balanced way. The expected level of formality is therefore
higher than many of the discourse types that the students engage in during spontaneous
spoken conversations. It is with respect to this difference that the term speech-
like/spoken register is used in this discussion as a short-hand for the more intricate
differentiations involved.
The frequency figures for modal verbs in the present study indicates that features of
the same tendency for spoken register can be found in L1 English learners of German.
We will see in the course of the further discussion that there is plenty more evidence for
this trend in various other modal aspects. This suggests that this feature of learner
language is not just L1, but also L2-independent.
In the lower proficiency year groups A and B, the overuse of modal verbs is, to some
extent, counterweighted by an underuse of other modal means (periphrastic modal
expressions, modal infinitives and Konjunktiv II). In the most advanced year group C,
however, Konjunktiv II forms and periphrastic modal expressions are also significantly
overused. This corroborates the findings of Hyland and Milton (1997), who report an
overuse of modal adverbials in learner writing, although their cross-sectional study only
pertains to epistemic modality. They show that advanced L2 English writers use more
epistemic devices overall than lower proficiency learners. Both Hyland and Milton and
239
the present study come to the conclusion that, while both lower and higher proficiency
groups rely heavily on certain modal verbs, it is the increased frequency of periphrastic
modal expressions, especially for epistemic modality, that characterises the most
advanced proficiency group. For learner German, we have additionally identified the
increased use of Konjunktiv II forms as contributing to the overuse of modal expressions
(this will be discussed in more detail in 6.2.8).
In summary, if all the different modal expressions are considered together, learners
exhibit a tendency to overuse modal expressions. This has been shown to tie in with a
more speech-like register, rich in modal verbs and periphrastic modal expressions. It
could, however, also be indicative of another general trend. The overuse of modal
expressions means that learners explicitly express modality more than native speakers.
This propensity for explicitness has also been suggested in other research as a potential
characterising feature of learner language. Lorenz (1999), for instance, investigates
adjective intensification in L1-German learners‟ writing in English. He concludes that
“learners are more anxious to make an impression than native speakers […] they employ
every means at their disposal […] the subjective effect is one of overstatement […] of
„too much communicative effort‟” (Lorenz 1999:199f) and “information overcharge” (ibid:
206). My data seems to corroborate these impressions. As soon as the learners have
certain means to express modality at their disposal, they employ them as much as
possible – to the point of „overcharge‟. This shows in the increased rate of periphrastic
modal expressions in the advanced year group, but also in the significant overuse of
Konjunktiv II forms in this group. This means that, unlike Lorenz (1999), we see the
tendency to employ explicit markers of modality not only in lexical, but also in
grammatical structures – another reason to assume a general feature of learner
language rather than a phenomenon tied in to one particular linguistic phenomenon.
It seems that, once the learners are confident that they have mastered this particular
category, they will endeavour to prove their proficiency by incorporating more and more
of these items into their writing. The fact that all the texts that have gone into the corpus
are part of the students‟ assessed coursework (where sometimes explicit „bonus marks‟
are awarded if a learner uses certain structures) surely is an influencing factor for the
learners‟ eagerness to make an impression.
240
6.2.2 Ranking of modal verbs
The previous section discussed, amongst other things, that the results of this study
clearly corroborate previous claims that modal verb frequencies are generally highly text-
type dependent. In this section, the focus is less on absolute frequencies, but rather on
the relative frequencies and rank order of modal verbs in the different corpora. The
comparison between the comparable native speaker corpus KEDS and the general
native speaker corpus LIMAS has uncovered no text-type specific difference, i.e. the
rank order of modal verbs in KEDS and LIMAS is, by and large, identical142:
KÖNNEN – MÜSSEN – SOLLEN – WOLLEN – MÖGEN – DÜRFEN
This is also identical to the overall rank order for the whole of the learner corpus
CLEG. If, however, the learner corpus is split into the different year groups, differences
can be observed. In the lower proficiency groups (Years A and B) the order is as follows:
KÖNNEN – MÜSSEN – WOLLEN – SOLLEN – MÖGEN – DÜRFEN
In the advanced proficiency group (Year C) the order is:
KÖNNEN – SOLLEN – MÜSSEN – WOLLEN – DÜRFEN – MÖGEN
The most striking difference is the prominence of WOLLEN over SOLLEN in the lower
groups. This prominence of WOLLEN is reminiscent of the order in Dittmar et al.‟s
studies (Dittmar 1979; Dittmar and Terborg 1991; Dittmar and Ahrenholz 1995) on the
emergence of modal verbs in natural settings, although the learners in Years A and B
are clearly more advanced than the subjects in their studies (see 3.1). Their findings
show that, despite some differences depending on the learners‟ L1, it is the modal verbs
WOLLEN, MÜSSEN and KÖNNEN which are acquired first, while SOLLEN, MÖGEN
and DÜRFEN are acquired much later. Dittmar (1979) identifies meaning extensions and
communicative needs as factors influencing this acquisitional order. With regard to the
instructed learners in the present study, the teaching material could also play a part.
Corpus evidence can, again, provide another possible factor for consideration. It has
been noted earlier that learners often exhibit a tendency to take on a more speech-like
register when writing in a foreign language. A comparison between spoken and written
corpus data from the DWDS corpus retrieving the relative frequencies of WOLLEN and
SOLLEN uncovered that WOLLEN is roughly twice as frequent as SOLLEN in the
142
In LIMAS, the last two, lowest frequency items MÖGEN and DÜRFEN are reversed.
241
spoken parts of the corpora. In written parts, the reverse is the case with SOLLEN being
twice as frequent as WOLLEN. This ratio is confirmed in the LIMAS corpus, the KEDS
corpus and also the advanced learner group subcorpus Year C. In the lower proficiency
groups, however, WOLLEN occurs up to one-and-a-half times as frequently as SOLLEN,
effectively making the relative frequencies and rankings of modal verbs more akin to
those of the spoken data. The prominence of WOLLEN in the CLEG data could therefore
yet again be ascribed to the more speech-like writing style of the lower proficiency year
groups, which, in turn, can be taken as a sign of the learners not mastering the L2 genre
conventions of argumentative writing, whilst the more advanced learners manage to
approximate more closely the appropriate register in this respect.
6.2.3 Form/meaning equivalents in learners‟ native and target language
Epistemic modal verbs with form/meaning equivalents in the target language and high
mutual correspondence values (MÜSSEN, KÖNNEN/KÖNNTE, WERDEN) have been
acquired by learners in all year groups and tend to be overused. Epistemic modal verbs
without form/meaning equivalents (evidential SOLLEN and WOLLEN, MÜSSTE,
DÜRFTE) or low mutual correspondence value (MÖGEN and SOLLTE) have not been
acquired even in the most advanced year group. Only SOLLTE shows some signs of
being acquired in this group. These findings support Lindemann‟s (1992) notion of a
„standard repertoire‟ (see 3.2) – consisting mainly of those modal verb meanings where
lexical transfer from the close cognates in the L1 is positive – and Giacalone Ramat‟s
(1999:390) statement that “the learner seems to look first for form/function units which
have a counterpart in the first language.” Other modal verb meanings are only acquired
at a later stage and with considerable difficulty, which is also in line with Lindemann‟s
(1992:144) analysis that transfer from the L1 is a key factor in modal verb acquisition.
Our study equally demonstrates that L1-influence certainly plays a decisive role in the
acquisition and use of modal verbs. In its most „visible‟ form, it occurs in the form of
negative transfer errors. The prominence of NICHT MÜSSEN instead of NICHT
DÜRFEN in order to express the notion of prohibition, especially in the lower proficiency
groups, is one case in point (see 5.1.2.2). The use of WERDEN in general event and
habitual contexts is another (see 5.1.1.6). In Jarvis and Pavlenko‟s (2008:75) definition,
these are cases of semantic transfer, as they involve “the use of an authentic target word
242
with a meaning that reflects influence from the semantic range of a corresponding word
in another language”. In English, the semantic range of the modal verb MUST, for
example, entails the notion of prohibition if the modal verb is negated and the learners
transfer this range into the target language143.
In the context of the language classroom such negative transfer is, of course,
classified and penalised as an error, but in the context of a description of learner
interlanguage, as is attempted here, these instances of transfer are valuable. They
indicate that learners utilise their knowledge of the L1 in order to produce and test
hypotheses about the L2. This is, of course, not a new idea, as the notion of transfer has
been the subject of second language acquisition research for many years.
However, the corpus data we have available now allows insights into transfer
phenomena that go beyond the observation of errors and include the analysis of patterns
of under- and overrepresentation, which in turn can shed new light on various factors
influencing transfer (cf. e.g. Jarvis and Pavlenko 2008:49). The close typological
proximity of German and English in general, and in the modal verb system in particular,
virtually invites the learner to transfer. For large parts of modal verb meanings the use of
form/meaning equivalents or close cognates (i.e. L1-transfer leads to a positive outcome
and thus facilitates learning). As these form/meaning equivalents also cover the most
prolific and frequent categories of epistemic and non-epistemic modality144, the learner
experiences a great number of receptive and productive instances, where a hypothesis
about the semantic and pragmatic characteristics of a German modal verb that is based
on the characteristics of the English cognate modal verb is confirmed. These modal
verbs therefore fulfil three crucial factors that Jarvis and Pavlenko (2008:177ff) identify
as interacting strongly with transfer: cross-linguistic similarity, frequency and
prototypicality.
Evidence from cognitive linguistics confirms that “L1 and L2 cognates have
exceptionally strong formal links in the bilingual mental lexicon, so that translation can
take place directly at the lexeme level without calling up the associated conceptual
representations” (Jarvis and Pavlenko 2008:85). In Jiang‟s (2002) model, form-meaning
mapping forms the essential part of vocabulary acquisition for L2 words with existing L1
143
Similarly, negative semantic transfer can be observed for some periphrastic modal expressions, e.g. the use of wenigstens in the sense of mindestens, where the English expression at least covers the semantic range of both German espressions (see 6.2.2.1).
144 These are: epistemic certainty (MÜSSEN, WERDEN) and possibility (KÖNNEN) and non-epistemic necessity (MÜSSEN), advice (SOLLTE) and possibility (KÖNNEN).
243
translations and concepts. He suggests that in the early stage of vocabulary acquisition,
L2 words are initially mapped to L1 translations, not to meaning directly:
“Each time an L2 word is encountered, its L1 translation is activated to provide
lemma information (i.e. meaning and syntax) […] With increased experience in L2,
which means increased coactivation of L2 words and their L1 lemma information, a
strong link is established between L2 words and the lemma component of their L1
translations. That is, L2 words are no longer mapped to L1 translations but to L1
meaning directly.” (Jiang 2002:619)
Jiang goes on to suggest that, once L1 semantic information has been copied into the L2
lexical entry, it becomes part of the lexical knowledge represented in L2 entries and it is
then very hard for new meanings to replace or add to the existing ones. The L1 semantic
information in the L2 lexical entry continues to mediate L2 word use even with continued
exposure to the L2. “As a result, even highly proficient L2 users will use L2 words on the
basis of the semantic specifications of their L1 translations” (Jiang 2002:619). The theory
that, once L1 semantic information has infiltrated the L2 lexical entry, it is hard to replace
or amend is supported by cognitive research that provides explanations for why
similarities and differences in L1 and L2 can influence acquisition. Ellis (2008) is able to
show that two relevant factors (proven to influence language learning) are proactive
inhibition and blocking. Proactive inhibition states that prior learning (e.g. the semantic
content of a word in the L1) inhibits new learning (e.g. the semantic content of a word in
the L2). Blocking means that once a certain cue is associated with a certain outcome
(e.g. a certain word with a certain semantic meaning), then the later association of
another cue with the same outcome is blocked due to the prior learning of the first cue
(and leads to automatically learned inattention).
All of these factors should contribute to the learners perceiving extensive similarities
in the modal systems of their native language and the target language145, and it is exactly
this deficiency in marked contrasts that, in turn, leads to overextensions of similarities.
This process is enforced even further by some learner textbooks and grammars (e.g.
Zeitgeist and Brennpunkt), which promote the impression of one-to-one equivalence
between many German and English modal verbs and therefore encourage transfer
without enough emphasis on the differences.
145
Although I agree with Lindemann (1992:146) that it is doubtful that learners possess declarative knowledge of the L1 that matches a linguistic description of the modal system as a whole.
244
At the same time, structures that are perceived by the L2 user as marked, i.e. as L2
specific or atypical, are less likely to transfer (cf. Jarvis and Pavlenko 2008:187). This
can be seen in the modal verbs without L1/L2 equivalents or low mutual correspondence
values. The use of modal verbs in order to express evidentials, for example, is a non-
prototypical use (although not exclusive to the German language). Here, the cross-
linguistic difference and markedness of these structures is paired with low frequencies in
the native speaker data. All of these factors discourage transfer and, in the case of my
learner data, lead to the avoidance or non-acquisition of these particular modal verb
meanings.
So far, we have seen that semantic transfer can explain certain error patterns and
avoidance. In order to explain the overuse of certain modal verbs, however, a different
type of cross-linguistic influence has to be considered: conceptual transfer. In line with
theories on linguistic relativity146 (i.e. the influence of language on thought) conceptual
transfer can be defined as those cases of linguistic relativity involving comprehension or
production in a second language (Odlin 2005:5). Unlike linguistic transfer (i.e. the
transfer of linguistic forms and structures), which entails e.g. semantic transfer as
described above, conceptual transfer relates to the transfer of the mental concepts that
underlie those forms and structures and consequently the “ways in which conceptual
representations are structured and mapped to language” (Jarvis and Pavlenko
2008:112).
According to Jarvis and Pavlenko (2008:113) children develop these conceptual
representations during the process of cognitive maturation. Conceptual development is
an experience-based developmental process that results in two types of conceptual
representations – those that are language independent and those that are language-
mediated. We are mainly interested in the latter category here, as this affects cross-
linguistic influence. Language-mediated concepts develop in the process of language
socialisation where word learning and category acquisition influence each other over an
extended period of time. Once children have internalised basic conceptual distinctions
characteristic of their first language, these language-mediated concepts then also guide
their attention to and selection of referential reality for subsequent memorisation and
description.
146
For a detailed summary of the ideas involved in the theory of linguistic relativity see e.g. Jarvis and Pavlenko (2008) and Odlin (2005).
245
As the focus in this study is on the perspective of cross-linguistic influence, we are
therefore concerned here with the effects that the language-mediated concepts of one
language have on the verbalisation of thoughts in another – conceptual transfer. It is this
type of transfer that can explain the extensive overuse of the modal verb WERDEN in
the learner data. The analysis established that it is WERDEN with future-time reference
that is the biggest contributor to the overuse by learners in all year groups. This can be
accounted for by the ways in which the concepts of present-time and futurity are
encoded in the learners‟ L1 and L2. In English, the simple present tense is unmarked
only for present-time events. It can be used for future events, but its use in this function
is highly restricted to future events that are perceived as absolutely certain, it is marked
and infrequent. In order to express future events, WILL and GOING TO are by far the
most widely applicable, frequent and prototypical forms. The present progressive is also
available where the future event results from a present plan or arrangement.
On the whole, any notion of futurity in English, be it immediate or not, has to be
expressed by some explicit marker and the learners seem to transfer the way the
concept of futurity is verbalised in their L1 to the L2 target language. However, all of the
different forms to express present and future time in English stand vis-à-vis just two
forms in German: the simple present and WERDEN. The difference is that, in German,
the present tense is available as a frequent and unmarked form for future events that
have some link to the present or where the utterance already contains some other
futurity marker (e.g. an adverbial). In English, these contexts require some form of future
marking. As WERDEN is the only available option in German, English learners of
German therefore feel compelled to use this form in contexts where German native
speakers often use the present tense. The overuse of WERDEN is therefore not simply
due to a semantic transfer of English WILL, but rather due to the fact that, in English, the
concept of futurity is much closer linked with explicit linguistic markers of future tense
than in German and it is this „need‟ for explicitness that is transferred.
The fact that there is no visible change in the learners‟ use of WERDEN with
increasing proficiency could also be taken as corroborating evidence that conceptual
representations are involved. Jarvis and Pavlenko (2008:171f) ascertain that the
internalisation of new concepts and restructuring of already acquired concepts are
lengthy processes dependent on many variables, which makes conceptual
representations far more resistant to change than links on a semantic level.
246
One particular finding in previous modality research might suggest an additional
reason for the overuse. Two studies on English as L2 with typologically very different L1s
(Cantonese and Swedish) both discovered that learners of English as well as novice
writers of English overuse WILL as their top favourite marker of epistemic assertion
compared to expert English L1 writers (Hyland and Milton 1997; Aijmer 2002). Neither of
these studies differentiates between WILL with present-time and future-time reference.
The learners in the CLEG corpus can be classed as novice writers in their L1 (see 4.2.2),
so that the conceptual transfer which leads to the overuse of WERDEN might be
reinforced by lexical transfer from WILL to WERDEN as a characteristic of novice writing.
It has to be stressed here that the observations and conclusions presented here as
evidence for conceptual transfer are only possible as a result of the type of quantitative
analyses that the corpus data allows for. As the subtle differences between the learners‟
and native speakers‟ use of this particular modal verb do not result in overt errors, it is
the diverging patterns of frequency that indicate „non-nativeness‟ and therefore lead to
further investigation. A qualitative study, such as Lindemann‟s (1996), does not permit
these kinds of discoveries.
Whilst we have seen strong links between form/meaning equivalents and usage or
avoidance patterns with the epistemic modal verbs, no such correspondences could be
observed for the non-epistemic modal verbs. Indeed, most non-epistemic modal verbs
are overused by the learners, regardless of form/meaning equivalents. This obviously
does not mean that L1-influence is not at work here, too. As with epistemic modal verbs,
the most frequent non-epistemic modal verbs (MÜSSEN, KÖNNEN/KÖNNTE, SOLLTE)
correspond largely in meaning and use to their English counterparts and cover a wide
range of modal meanings. The learners, again, experience a vast amount of instances
where L1-transfer serves them well and there are no reasons why transfer should not
occur here. On top of this, teaching and learning materials (such as the aforementioned
textbooks Brennpunkt and Zeitgeist) also often promote easy one-to-one relationships
for the other non-epistemic modal verb meanings, e.g. DÜRFEN = be allowed to,
WOLLEN = want etc. All of this facilitates the acquisition and successful application of
non-epistemic modal verbs by learners even at an early stage.
The reasons for the consistent overuse of non-epistemic modal verbs, however, have
to be sought elsewhere. Factors such as a more speech-like register in learner language
or the tendency for explicitness, which have been discussed in the previous section,
247
seem to be the most plausible e.g. for explaining the consistent overuse of MÜSSEN to
express non-epistemic necessity and WOLLEN, MÖCHTE, MÖGEN to express volition
and liking/affection.
6.2.4 Correspondences between usage patterns of modal verbs and periphrastic modal expressions
In the previous section, we discussed how form/meaning equivalents influence the use of
epistemic modal verbs, but not those of non-epistemic modal verbs. One observation
that applies to both epistemic and non-epistemic modality alike is correspondences
between the usage patterns of modal verbs and those of their corresponding periphrastic
modal expressions.
The non-acquisition of certain epistemic modal verbs (MÜSSTE, SOLLTE, DÜRFTE,
MÖGEN) leaves semantic gaps for which the learners compensate with the overuse of
periphrastic modal expressions in order to express the intended meanings of mitigated
certainty (e.g. ich denke, meiner Meinung nach etc.) or possibility (e.g. vielleicht, es ist
möglich etc.).
Although there is no corresponding pattern of modal verb underuse in the non-
epistemic categories, the learners still tend to underuse non-epistemic periphrastic
modal expressions where the corresponding modal verb has an L1 form/meaning
equivalent (MÜSSEN, SOLLEN/SOLLTE, KÖNNEN/KÖNNTE). This means that they
tend to use periphrastic modal expressions denoting a duty or obligation, such as es ist
notwendig less than the native speakers, as the modal verb MÜSSEN supplies the same
meaning and is at the same time the form/meaning equivalent to English MUST. At the
same time, the learners tend to overuse periphrastic modal expressions for those
meanings where no corresponding form/meaning equivalent exists for the modal verb
(WOLLEN, MÖCHTE, MÖGEN, DÜRFEN). This means that, for instance, expressions of
wishes such as hoffen are overused by the learners, irrespective of a simultaneous
overuse of the modal verb WOLLEN.
This suggests that, even where there is no semantic gap in learners‟ knowledge, they
tend to employ the type of modal means that would be used in their native language
rather than in the L2. The concepts of volition, liking and permission are expressed by
periphrastic modal expressions in English, therefore learners also often express these
meanings by periphrastic modal expressions when writing in German. This indicates a
248
type of transfer that Jarvis and Pavlenko (2008:88) have termed „word choice transfer‟,
as it relates to the “preferential selection of certain words over others in contexts where
there exist multiple alternatives”. This type of transfer also pertains to other linguistic
structures or grammatical categories, for example the choice of an adverbial over a
modal verb. Several studies have confirmed that the choice of a specific L2 word,
structure or grammatical category in a specific context is often motivated by a
corresponding L1 preference (e.g. Jarvis and Odlin 2000; Jarvis 2000; Pavlenko and
Driangina 2007).
But, at the same time, the learners also follow an L2-specific trend that has been
noted in the literature (see 2.1.3) and is also attested in my native speaker corpus data,
namely that, in German, periphrastic modal expressions are preferred for epistemic
modality both in spoken and written discourse, while for non-epistemic modality, modal
verbs are the preferred option. The prominence of periphrastic modal expressions over
modal verbs in the learners‟ German production clearly contradicts the tendency in their
native language, where epistemic modality is prototypically and most frequently
expressed by modal verbs (cf. e.g. Aijmer 2002: 66)147.
Hyland and Milton (1997) argue that the reason why periphrastic modal expressions,
especially adverbs, are so popular with learners is that they are easy for novice writers to
manipulate: they are less syntactically integrated into the clause structure than modal
verbs and more mobile in terms of sentence position. Whilst this argument is certainly
also applicable to German, there are additional factors to be considered, in particular for
the epistemic category. The prominence of periphrastic modal expressions in epistemic
meanings means that they are therefore much more salient for the learners than
epistemic modal verbs, as the learners encounter periphrastic modal expressions in
epistemic functions frequently in the target language. (We adopt here Bardovi-Harlig‟s
(1987:387) definition of salience as “availability of data”.) As epistemic meanings of
modal verbs are infrequent (both in terms of absolute frequency and relative to the non-
epistemic meanings), and as there is often a possibility of ambiguity between epistemic
and non-epistemic modal verbs, it seems fair to assume that it is difficult for the learners
to pick up these particular meanings and extract the correct grammatical and pragmatic
use of these modal verbs from the very limited input.
147
This tendency is also followed by L2 learners of English, who overall use epistemic modal verbs more frequently than modal adjectives or adverbs (cf. e.g. Hyland and Milton 1997; Gabrielatos and McEnery 2005).
249
The argument that salience and „easiness‟ play a role in the preference for epistemic
periphrastic modal expressions by learners of German is supported by the fact that
classroom learners follow the same path as learners in natural acquisition settings here:
Propositional attitude verbs and modal adverbs precede modal verbs in epistemic
contexts in the acquisitional order irrespective of the target language (Giacalone Ramat
1999; Dittmar and Ahrenholz 1995; see 3.1). This is further supported by the fact that the
learners in my study prefer modal verbs over periphrastic modal expressions for the
expression of non-epistemic modality, just as the natural acquisition learners.
Yet another noteworthy aspect arises from the comparison between novice writers in
the learner corpus (CLEG) and comparable native speaker corpus (KEDS) and expert
native speaker writers in LIMAS. The modal adverbial vielleicht (perhaps) is highly
overused by learners of German compared to the similar native speaker group (in the
KEDS corpus). A comparison with a similar English native speaker corpus (LOCNESS)
ruled out L1-influence. In turn, vielleicht is also overused by the novice writers in the
comparable native speaker corpus KEDS in comparison to expert writers in the general
native speaker corpus LIMAS. This suggests that particular usage patterns of certain
modal expressions could also be a reflection of the difference between novice writers
and expert writers in general, irrespective of whether they write in their native or a foreign
language.
6.2.5 Difficulties with morphological and semantic distinction of indicative and Konjunktiv II modal verb forms
Learners in all year groups exhibit ongoing difficulties with the Konjunktiv II forms of
modal verbs (MÜSSTE, SOLLTE, KÖNNTE, DÜRFTE). These manifest themselves in
the avoidance148 of these modal verbs and in errors. The character of the errors is such
that the learners are unable to semantically distinguish between the different meanings
of the indicative and Konjunktiv II modal verb forms.
This happens in epistemic as well as non-epistemic modality. For example, learners in
Year A erroneously use MÜSSEN instead of MÜSSTE in order to express a tentative
epistemic conclusion. In Years B and C, MÜSSTE is also avoided, although there are no
148
Augustin (2006:75) also found in her analysis of Russian learners of German that the subjects avoided modal verbs in their subjunctive form. She ascribes this to the fact that there are substantial differences in the formation and usage of Russian and German subjunctive forms.
250
incorrect uses of MÜSSEN anymore. In non-epistemic modality, MÜSSTE is also
avoided until Year C where it is only used in hypothetical or conditional contexts, i.e.
contexts that require MÜSSEN in its Konjunktiv II form for grammatical reasons, rather
than in contexts that denote the semantic category of „advice/suggestion‟ (where its
meaning is synonymous to that of SOLLTE). Similarly, SOLLTE is avoided in Years A
and B in epistemic modality and the occurrence of several instances of SOLLEN instead
of SOLLTE in Year C points towards the same confusion as with epistemic MÜSSTE. A
closer look at SOLLEN in non-epistemic modality revealed that learners in all year
groups use SOLLEN instead of SOLLTE extensively to cover the semantic meaning of
„advice/suggestion‟, although SOLLEN actually denotes the category
„instruction/obligation‟, which is on a par with MÜSSEN in terms of modal force. Equally,
problems are noticable with KÖNNTE, although here errors are only visible in incorrect
uses of the preterite form KONNTE in contexts where KÖNNTE would be required and
vice versa.
These results have two implications. First of all, the reason why learners avoid these
modal verbs in their epistemic meanings might not just be to do with the non-existence of
L1 form/meaning equivalents (i.e. transfer issues) as discussed in 6.2.3, but also with the
difficulties that the semantic distinction between the Konjunktiv II and indicative forms
pose to the learners.
Secondly, the fact that learners have problems with the Konjunktiv II modal verb forms
in both epistemic and non-epistemic meanings indicates that the root of the semantic
problem is morphological. The morphological make-up of the German Konjunktiv II
modal verbs creates a two-fold problem for the learner. On the one hand, more
morphological forms have to be distinguished in German in comparison to English and
on the other hand, anecdotal evidence points to the fact that the learners‟ difficulties can
be traced to a general failure in phonological discrimination.
Where morphological distinctions are concerned, English modal verbs have only one
morphological indicative form and one form that can serve as the past tense indicative
form or as a form that is different from the modal meaning of the corresponding indicative
form149, i.e. CAN – COULD, WILL – WOULD, MAY – MIGHT, SHALL - SHOULD. There
is no corresponding form for MUST. In German, the preterite and Konjunktiv II modal
149
Nehls (1986:158ff) gives a detailed account of the differences between the past tense and modal meanings of the modal verbs COULD, WOULD, MIGHT and SHOULD. He stresses that in most cases and especially in epistemic meanings, these forms do not denote past-time reference but a specific modal meaning with present-time reference.
251
verb forms of KÖNNEN, MÜSSEN, MÖGEN and DÜRFEN are morphologically distinct,
while they are identical for SOLLEN and WOLLEN (see 2.3.1.1). Learners therefore have
to choose between different morphological forms for past-time reference and Konjunktiv
II modal meanings in the target language (but only for some of the modal verbs), where
these are all represented by the same morphological verb form in their L1.
This kind of difference between the L1 and the L2, where the native language has one
form and the target language has more than one (called „differentiation‟) has been
identified as one of the most difficult to grasp for the learners (Gass and Selinker
2008:100). The results from this study clearly indicate that this causes a major problem
in the acquisition of German modal verbs. As the learners are unsure about the different
morphological forms of past tense and Konjunktiv II in contrast to the indicative forms, it
is only a logical consequence that they also have difficulties in the semantic distinctions
between e.g. the indicative present tense form and the Konjunktiv II form. So, if learners
use SOLLEN instead of SOLLTE in contexts of „advice/suggestion‟, it is the underlying
failure to differentiate between the two forms and to then map the two distinct forms onto
different semantic meanings. It could be speculated that the learners might not even be
aware of the fact that these verb forms are Konjunktiv II forms of the modal verbs and
that they are connected by systematic grammatical regulations.
The way in which the different forms are often presented in textbooks and learner
grammars does its part to further obscure these distinctions and actually make it harder
for the learners to even notice them. In the reference grammar recommended to the
students (see Footnote 90), Hammer‟s German Grammar and Usage (Durrell 2002), for
example, the different uses of the Konjunktiv II modal verb DÜRFTE are subsumed
under the heading “DÜRFEN”, where the introductory text mentions that DÜRFTE is the
Konjunktiv II form, but in the subsequent examples, no clear indication of this is given.
Under the heading “KÖNNEN”, numerous examples of KÖNNTE are given, without even
mentioning that this is the Konjunktiv II form. No examples of or differentiations from the
preterite forms DURFTE or KONNTE are given.
Another factor that makes the differentiation between the various morphological forms
problematic for the learners is the fact that the phonological (and graphical, for that
matter) make-up of these forms also entails forms that exist in the learners‟ L2 but not in
their L1. Both umlauts that are involved in the modal verbs – ü and ö – are phonemes in
German: [y] and [œ]. Neither of these vowel sounds exists, let alone represents a
phoneme in English. The author knows from her own teaching experience that learners
252
very often fail to perceive the difference between, particularly, [u] and [y]. This is easily
explicable as these sounds do not constitute different phonemes in English. Ellis
(2008:392) documents that “a sound difference that crosses the boundary between
phonemes in a language is more discriminable to speaker of that language than to
speakers of a language in which the sound difference does not cross a phonemic
boundary”. This means, that if the learners perceive no difference between the two
sounds, the two sounds cannot function as separate stimuli that trigger different
interpretation outcomes. The difference is simply not obvious enough for the learners, so
they do not experience separate stimuli when presented with the two sounds.
In turn, if the learners cannot differentiate phonologically between the two sounds that
are central to the morphological distinction, it is not surprising that they have great
difficulty in not only separating the different verb forms but subsequently also their
different meanings. In other words, the failure to notice the phonological differences
leads to failure to semantically differentiate and inhibits learning. This is captured in
Schmidt‟s (e.g. 1990) “noticing hypothesis” and in Gass‟ (1988) notions of “apperceived
input”, which both suggest, in essence, that in order to turn input [the L2 data available in
the learner‟s environment] into intake [the subset of the input that the learner
appropriates to build the interlanguage] and make it available for further processing,
relevant input features have to be noticed, meaning they have to be registered under
attention, or detected under awareness (cf. Schmidt 1995; Schmidt 2001). It is generally
recognised that this can only be achieved by focussing the learner‟s attention specifically
to these features and that problems can only be remedied by making them salient to the
learners with the help of selected input (see e.g. Sharwood Smith 1981; Terrel 1991;
Doughty and Williams 1998).
With a specific view to the present study, learners would benefit greatly from any
measures that helps improve their grasp of the particular phonological differences
between vowels and umlauted vowels and their impact on morphological forms as this
has consequences for a wide range of lexical and grammatical structures. My analysis
has shown that, on top of the problems with the modal verbs, it also affects the
production of the Konjunktiv II, where the distinction between WERDEN, WURDE and
WÜRDE is crucial. Even when it does not result in outright grammatical errors, the failure
to distinguish e.g. WERDEN and WÜRDE has bearings on the rhetorical impact of an
argument, when e.g. an utterance is marked as a confident assumption (WERDEN),
253
where a more tentative, hypothetical note (WÜRDE) would be more appropriate,
especially if the surrounding context is all phrased using Konjunktiv II forms.
254
6.2.6 Personalised expressions
Learners show a distinct preference for personalised expressions compared to native
speakers. Personalised expressions “explicitly involve the writer in the assessment of
propositional validity” (Hyland and Milton 1997:197), typically by use of a first person
pronoun e.g. in expressions such as ich denke (I think) or meiner Meinung nach (in my
opinion). “Impersonalised forms, on the other hand, avoid reference to the writer when
commenting on the truth of a claim and typically conceal the source of epistemic
judgements” (Hyland 1997:197). This is typically achieved by adverbs such as
wahrscheinlich (probably), impersonal pronouns, e.g. es ist möglich (it is possible) or
passive constructions.
The learners‟ preference for personalised expressions is most clearly manifest in the
highly significant overuse of the periphrastic modal expressions category „subjective
assumption‟ in the lower proficiency year groups (Years A and B). In particular, the items
ich denke, ich glaube and meiner Meinung nach/ich bin der Meinung feature prominently.
At the same time, the category that contains impersonalised expressions that denote
roughly the same degree of epistemic certainty – „tentative conclusions‟ – are underused
in Year A and not overused in Years B and C. And even in this category, the propensity
for personalised expressions shows. The most frequent item in all learner groups for this
category is scheinen (seem), which can be used for an impersonalised expression – es
scheint (it seems) – but it can also be turned into a personalised expression by inclusion
of a first person pronoun – mir scheint/es scheint mir (it seems to me). In Year A, the
learners use scheinen in a personalised form in 54% of cases, in Year B it is still 20%.
Scheinen is not the most frequent item in KEDS anyway (instead the native speakers
prefer the impersonal adverb wohl (probably)), and when it does occur, it is always in an
impersonalised form.
Other categories of modal verbs and periphrastic modal expressions provide further
evidence for the same trend for personalised expressions. In the category for compelling
conclusions, the item ich bin überzeugt (I am convinced) figures much more prominently
than in the native speaker group, although the category as a whole is significantly
underused by all learner groups.
Unlike native speakers, learners also do not refrain from using positive evaluative
expressions, such as zum Glück/glücklicherweise (luckily) or Gott sei Dank (thank God)
255
or expressions of volition, such as ich hoffe (I hope) and hoffentlich (hopefully), all of
which indicate writer involvement.
Another aspect that is linked to personalised expression is the use of explicit agentive
subjects in non-epistemic modality. Although these are not necessarily first-person
subjects and therefore do not indicate the writer‟s own personal involvement, they
indicate that the writer assigns the „responsibility‟ for a certain modal meaning to a
specific animate agent. The analysis clearly shows this for the modal verb SOLLTE in
the meaning of „advice/suggestion‟. The native speakers prefer impersonal constructions
(passive voice, „man‟ (one) or dummy subject „es‟ (it)), which express a general
suggestion that the proposition be carried out, but refrain from assigning the
responsibility for this to any particular agent. The learners, on the other hand, prefer
SOLLTE with specified agentive subjects that explicitly identify the person or group that
is to take the proposed action.
Other categories where this tendency can be demonstrated are „liking/affection‟ and
„volition‟. In these two categories, it is the general overuse of these two categories that
indicates the greater propensity for learners to use specified agentive subjects. While the
native speakers tend to avoid utterances that express positive likings or volitions of
subjects, the learners overuse both modal verbs (MÖGEN, WOLLEN/MÖCHTE) and
corresponding periphrastic lexical expressions (e.g. lieben - love, hoffen - hope,
wünschen - wish) to do exactly that150. The overuse of agentive subjects is one finding
that has not been mentioned in the literature as a characteristic of learner language.
Further comparisons with other learner groups and with other languages could determine
whether this trend is universal.
It has to be added that the preference for personalised expressions and specified
active agents is strongest in the two lower proficiency groups and diminishes greatly in
the advanced group. In this year group, the personalised expression ich hoffe (I hope) is
replaced by the impersonalised expression es bleibt zu hoffen (it remains to be hoped).
Also depersonalised forms are used (where the view is attributed to another person, and
therefore responsibility is shifted), in order to escape the personalisation trap; but – much
the same as in English (Granger 1994 cited in Hyland and Milton 1997) – learners rely
150
However, in the particular category of „liking‟ a possible L1-influence should not be discounted, as an investigation into the corpus of native speaker argumentative essays (LOCNESS) also revealed an overuse of the word „like‟ in the sense of the German modal verb MÖGEN compared to KEDS.
256
heavily on sagen (e.g. viele Leute sagen – many people say), which is significantly
overused by all learners compared to native speakers151.
Personalised expressions could therefore be considered as an indicator of L2
proficiency, especially as this feature of learner language seems to be language-
independent. Several studies on learner English also demonstrated a significant overuse
of I think by learners of different L1 backgrounds (Granger 1998b; Ringbom 1998;
McCrostie 2008) as well as a general tendency of non-native speaker groups to overuse
features of writer visibility or involvement features (Hyland and Milton 1997; Petch-Tyson
1998; Aijmer 2002, Ädel 2008).
Again, this feature is associated more with spoken discourse than with written, so its
overuse by the learners, once again, points to a more speech-like register especially at a
lower proficiency stage. At the same time, the novice writers in my comparable corpus
KEDS also exhibit a tendency to overuse items like ich denke in comparison to the
expert writers in the general native speaker corpus LIMAS. While it can be considered an
indicator of L2 proficiency, it can therefore at the same time highlight “novice writers‟
uncertainty in how to use these items effectively in argumentation and their imperfect
grasp of appropriate register” (Hyland and Milton 1997:192).
6.2.7 Range of modal expressions and formulaic sequences
In the use of periphrastic modal expressions, learners continuously expand their
vocabulary range in all categories over the four years of study from lower to more
advanced proficiency groups. This is, of course, to be expected, as one of the indicators
of growing proficiency is an increased command of lexical items. But despite this
achievement even the most advanced learner group still exhibits a smaller range of
items than the comparable native speaker group. In Year C, the learners manage to use
about 75% of the different periphrastic modal expression types that occur in KEDS.
Moreover, a closer look at the frequency and distribution of individual items reveals
that many of them are only used sparsely by a handful of learners, with the majority of
tokens concentrated in a much smaller set of items than the native speakers. The same
development can be observed for the Konjunktiv II on lexical verbs. There is a steady
151
Raw frequencies: Year A: 77, Year B: 233, Year C: 168; KEDS: 43; Frequencies per 10,000 words: Year A: 15.7***, Year B: 35.4***, Year C: 19.2***, KEDS: 5.2 (*** = significant at p<0.001 compared to KEDS).
257
increase in the range of lexical verbs that are used in Konjunktiv II forms from Year A to
Year C, but the breadth of this range does not quite reach the native speakers‟ level
(88%) and is mainly restricted to a few items. The Konjunktiv II will be discussed in more
detail in the next section. What is of relevance in this section is that the increased (but
still comparatively small) range goes hand in hand with an increase in overall frequency
of both periphrastic modal expressions and Konjunktiv II forms, to the point where both
are overused in Year C compared to KEDS. This means that the learners, especially in
the highest proficiency group show a general propensity to overuse a smaller set of
items, i.e. they repeatedly use the same expressions.
Our analysis has shown that many of these expressions develop into fossilised
formulaic sequences during the course of the learners‟ progression. I adopt here Wray‟s
(2002:9) definition of the formulaic sequence as:
“a sequence, continuous or discontinuous, of words or other elements, which is, or
appears to be, prefabricated: that is, stored and retrieved whole from memory at
the time of use, rather than being subject to generation or analysis by the language
grammar.”
Their main function in general, i.e. for native speakers and learners alike, is identified as
serving “a single goal: the promotion of the speaker‟s interests” (Wray 2002:95). These
interests include, amongst others, having easy access to information and expressing
information fluently through the reduction of processing effort by using chunks that are
stored and retrieved whole; being listened to and taken seriously through easing the
processing effort for the listener; sounding idiomatic and being perceived as a full
member of whichever groups are deemed desirable.
Wray (2002) develops a model of formulaic sequences within the second language
acquisition process on the basis of a focused review of data provided by a multitude of
other studies. Within this review, she observes a clear distinction between „naturalistic‟
learners and (teenage and adult) classroom L2 learners. The latter is relevant for my
study and the following points characterise the nature and use of formulaic sequences
for classroom learners:
While formulas are seen as an „easy‟ way of meeting communicative needs (e.g.
greeting, apologising etc.), especially for beginning learners, the communicative
ambitions of advancing learners soon outstrip the limitations of fixed expressions and
258
creativity becomes a more central issue. This produces the following observed
progression: Beginners in the earliest stages use a very limited set of formulaic
sequences extensively. In later stages, the successful learners will retain formulaic
sequences but also expand their creative use of language. For intermediate and
advanced learners, however, who have acquired a reasonable command of the L2
lexicon and grammar, their command of formulaic sequences appears to be lagging
behind their other abilities. Full presence of formulaic sequences is only observed in
learners who have achieved a near-native level. The difficulty for intermediate and even
advanced learners seems to be to strike the right balance between formulaicity and
creativity in their L2 output.
Several studies observe overuse AND underuse patterns of formulaic sequences as
well as too much creativity within them, i.e. learners make creative changes to
sequences that native speakers use as fixed expressions (e.g. Granger 1998b; DeCock
1998). This leads to the impression that even “advanced learners struggle to sound
idiomatic, even when they sound grammatical” (Wray 2002:191). But learners‟ formulaic
sequences also often contain errors. The question is, if they are stored and retrieved as
whole items, how can these errors occur? On the one hand, the data in the reviewed
studies shows memorised chunks that are reproduced whole in contexts where they are
incorrect. On the other hand, there is ample evidence that learners create their own
chunks based on their interlanguage rules. If their assumptions about these rules are
incorrect, the chunks that learners produce may contain errors that often fossilise into
persistently problematic patterns. But it is exactly this evidence, that learners create their
own sequences, which marks the difference between (classroom) second language
acquisition and first language acquisition in Wray‟s model (2002:210).
Wray postulates that, in first language acquisition, formulaic sequences are, indeed,
stored and retrieved whole. Analytical processes only set in if and as much as is
necessary. In instructed second language acquisition, on the other hand, learners do
apply analytical processes to the formulaic sequences they encounter. The reasons for
this lie in the very different learning conditions for child first language learners and
teenage/adult classroom second language learners and in their different intellectual
experiences (e.g. literacy) and cognitive maturation. This means “L2 learners will fall into
the process of analysis which provides a more manageable outcome and offers a
stronger sense of control over the language material” (Wray 2002:212). Classroom
teaching methods characteristically reinforce the analytical approach, e.g. through „focus
259
on form‟ and reliance on written materials, which underlines the importance of small units
(individual words) over bigger sequences. Fundamentally, learners have difficulties with
formulaic sequences (whose purpose is, after all, to make language use easier),
because they have too much lexical and grammatical choice, as learners build up
sequences from words, while native speakers possess them as whole sequences and
break them down only if necessary.
Evidence for both aspects in Wray‟s account – the path of progression and the
analytic nature of formulaic sequences in L2 learners – can be found in the present
study.
When looking at the way certain items in the observed repertoire of periphrastic modal
expressions are encapsulated in formulaic sequences, we can see a distinct progression
from analytically produced chunks to fossilised sequences. The noun Tatsache (fact)
shall serve as an example here. In the lower proficiency years, the noun is typically
syntactically integrated into the sentence structure in a variety of ways that show analytic
processes in their production152, e.g. trotz der Tatsache (despite the fact); Die Tatsache,
dass…, beweist… (The fact that… proves…); Ein großes Problem liegt *mit der
Tatsache... (a big problem lies with the fact…). These are actually the kinds of structures
that native speakers also prefer. In the most advanced year group, however, the majority
of uses of Tatsache occurs within the fossilised formulaic sequence Es ist eine Tatsache,
dass… (It is a fact that…). While not grammatically wrong, it is clearly a structure that is
dispreferred by the native speakers.
There is an abundance of evidence in this advanced year group of similar fossilised
formulaic sequences, which are not only different from the native speakers‟ uses of the
respective modal expressions within them but also greatly overused by the learners.
Moreover, as noted above, they operate on a smaller range of lexical items than the
native speakers, with some „favourites‟, such as ich denke or meiner Meinung nach,
occurring in significantly higher frequencies. This finding is in line with Granger (1998),
who suggests that learners tend to settle on a small number of familiar and safe
sequences which become “islands of reliability” (Dechert 1983). Granger specifically
identifies the “learners‟ repertoires for introducing arguments and points of view” (which
modal expressions are a big part of) as “very restricted”, which is why learners “„cling on‟
to certain fixed phrases and expressions which they feel confident in using” (Granger
152
Some of which involve L1 interference.
260
1998:156). This is similar to Hasselgren‟s (1994) notion of „lexical teddy-bears‟. Native
speakers, on the other hand, use a greater variety of expressions and, with them,
produce a greater variety of chunks. Again, my results can be viewed as corroborating
evidence for a language independent feature of learner language.
Another striking aspect is the uniformity that the learners‟ sequences exhibit where
the various lexical items from the repertoire of modal expressions are concerned. The
formulaic sequences in my learner data typically display the following distinct
characteristics that set them apart from the native speaker sequences:
They are syntactically extrapolated into a matrix clause
They are sentence-initial, i.e. thematised and focused elements of the utterance
They are often of the type Es ist…, dass… ( e.g. Es ist möglich, dass...; Es ist
nicht zu leugnen, dass...; Es ist eine Tatsache, dass... etc.) or a variation with a
similar verb (e.g. Es gibt die Möglichkeit, dass...; Es gibt keinen Zweifel, dass...;
Es scheint, dass...; Es bleibt zu hoffen, dass... etc.)
Taking into account Wray‟s proposition that learners create their own chunks from
individual words rather than simply taking over prefabricated sequences they encounter
in the L2, it is not surprising that their formulaic sequences look different from the native
speakers‟. A reminder of previously established tendencies in learner language can
provide answers to the question of why they look the way they do. As noted before
(cf.6.2.4), periphrastic modal expressions are less syntactically integrated in the clause
(i.e. they lend themselves to extrapolation), which reduces the processing strain that an
integrated structure poses on the learner153. Extrapolating the formulaic sequence,
therefore, allows the learner to separate syntactically (and therefore possibly also
mentally) the indicator of modality from the proposition they wish to modify.
With regard to the sentence-initial position of the sequences, we can also point to the
separation of modality marker and proposition for an explanation. In addition to this, the
sentence structure follows the default pattern of information structure in both English and
German: “The structural differences between English and German are not so great as to
entail a fundamentally different way of structuring information […] in both languages the
given information should provide the point of departure of a message, and […] this
shared point of reference should be followed by what is „news‟ to the reader” (Lorenz
153
cf. also the conclusions of Kärkäinnen (1992) (summarised in section 3.1) that syntactically integrated, non-routinised expressions of epistemic modality (such as modal verbs and adverbs) – which convey speaker attitudes implicitly – are more difficult to acquire than explicit, extra-clausal and routinised expressions, such as parentheticals.
261
1999:211). The aim of argumentative texts is to convince the reader of the writer‟s way of
thinking. Therefore, presenting the part of an utterance that contains an indication of the
writer‟s opinion as the „shared point of reference‟ makes it easier for the reader to accept
and take over the writer‟s stance towards the proposition (Schmied 2001).
Another explanation could be that these sequences are attempts of the learners to
strive towards impersonalised expressions. We have seen in 0 that learners in the lower
proficiency groups tend towards personlised expressions and that it is only in the higher
proficiency group that learners manage to revoke this tendency as they gradually learn
that personalised expressions are less register and text-type appropriate (showing in a
significant reduction in the frequency of the „subjective assumptions‟ category, for
example). The impersonal construction with the dummy subject es offers a way out of
this which is not only convenient but also highly flexible with respect to the lexical content
it can be filled with. Again, research on learner of English confirms that this strategy
seems to be universal and not language specific. Schachter and Rutherford (1979) and
Han (2000) report that Chinese and Japanese learners of English exhibit a distinct
overuse of so called topicalisation constructions, such as there is and it is. They attribute
this to learners trying to follow the topic-complement template of their L1s, even though
their L1s do not have structures directly comparable to there is and it is (cf. Jarvis and
Pavlenko 2008:51). The evidence from my study indicates, however, that L1-influence
might be only a minor factor in this and that the more prominent factors are, actually, not
transfer related. Finally, as all of these explanations point towards learner- and
language-independent factors, they can also serve to explain why the same fossilised
patterns are produced by so many different learners.
There is one final aspect where the results of my study provide corroborating
evidence for Wray‟s model of formulaic sequences of language learners, and this
involves the modal infinitives. My analysis has shown that learners largely avoid modal
infinitives and do not seem to possess any meta-linguistic knowledge of the grammatical
category „modal infinitives‟154. Where they do occur, they are usually part of formulaic
sequences in the learner specific form that has been discussed in this section (e.g. Es ist
nicht zu leugnen…, Es ist leicht zu sehen… etc.). This presents us with the following
conundrum: If, as Wray proposes, learners construct their own formulaic sequences by
154
Which shows e.g. in the failure of students to realise the potential for ambiguity in modal infinitives.
262
involving analytical processes, how can they produce modal infinitive constructions when
they do not possess any meta-knowledge of their general grammatical structure. Or, if
the question is turned around, “why do adult learners fail to extract grammatical
information from formulaic sequences?” (Wray 2002:210). Again, Wray provides a
compelling explanation. She proposes that although some kind of analysis is happening
when learners encounter formulaic sequences, this analysis seems to be ignoring
grammatical information. Instead,
“the characteristic way for formulaic sequences to be handled by post-childhood
language learners is that they are broken down in order to access the lexical
constituents, which are then stored separately. […] how they were linked,
however, is not retained. […] This also means that when the learner wishes to
reconstruct the formulaic sequence it will be necessary to apply the interlanguage
grammar” (Wray 2002:210) [my emphasis].
This is how errors occur in formulaic sequences. With the sequences involving modal
infinitives, there are clear indications that the learners‟ interlanguage grammars rely
largely on cognate L1 structures which are transferred, rather than on rules for
constructing modal infinitives. This applies especially to the prominent sequences of the
type Es ist leicht/schwer zu + SENSORY PERCEPTION VERB…. The way native speakers
prefer to use modal infinitives, i.e. with mental/speech act verbs or physical action verbs,
is not realised by the learners.
Teaching materials provide another piece of evidence that learners analyse the
formulaic expressions they encounter for lexical constituents but not for their
grammatical construction. Writing critical commentaries and other argumentative texts
are recurring tasks that form part of the students‟ assessment in all three years of study.
How to write a good argumentative essay is therefore also a regularly recurring topic of
the language classes and the students received appropriate „Redemittel‟ (useful
phrases) and exercises to improve their repertoire of phrases as part of these classes (in
the absence of a set textbook, these were provided in the form of handouts put together
from various resources by the teacher; for examples see Appendix D, Part I). From these
handouts, we can see that they contain the type of formulaic sequences that are deemed
useful for learners to facilitate introducing arguments and counter-arguments and make
clear the writer‟s position towards a proposition – in other words formulaic sequences
that denote epistemic modality. These sequences, however, are not the ones that
263
learners actually use. As we can see, and as predicted by Wray‟s model, the learners
extract from them some lexical components, but they do not store the grammatical
structure of these formulas: they construct their own sequences using their interlanguage
grammars. In the lower proficiency years, this sometimes results in errors, e.g. in
sequences like *in meiner Meinung (L1-transfer of in my opinion). In the advanced group,
where these errors are largely avoided, it leads to fossilised sequences of the Es ist…
kind. While grammatically correct, their overabundant occurrence, together with their
specific structural properties, leave the reader with a distinct impression of „non-
nativeness‟.
This is, of course, exactly the opposite of the desired effect, namely to approximate a
text-type appropriate writing style, e.g. by avoiding personlised expressions, and
promote the quality of the text by including what the learners perceive as „useful‟
formulaic phrases. This is, after all, what they are told to do in their language classes,
especially in the current prevailing teaching practices which “positively reward such
memorised material” (Mitchell and Martin 1997:6). Additionally, we can see that some of
the Redemittel (useful phrases) that the students are given in the context of verbal
discussions (see Appendix D, Part II) find their way into the written texts in unusually
high frequencies – another pointer towards a speech-like register.
6.2.8 Konjunktiv II as indicator of learner proficiency
The usage patterns of the Konjunktiv II that have been uncovered in my analysis suggest
that the Konjunktiv II can serve as an indicator of learner proficiency, as a marked
progression can be observed both in terms of overall frequency of the Konjunktiv II and
in the diversity of lexical verbs used in the Konjunktiv II. We have seen in previous points
that some aspects of the learners‟ use of the Konjunktiv II fit in with general observations
of features of learner language. These relate to the tendency of learners to explicitly
mark modality (see 6.2.1), specific error patterns (see 6.2.5) and an expanding lexical
range (see 6.2.7). In this section, we will concentrate more specifically on how these
aspects affect the Konjunktiv II usage patterns and what implications this has for learner
proficiency.
Learners significantly underuse Konjunktiv II in the first two years (Year A and B). In
the advanced group (Year C) it is significantly overused compared to the native speaker
264
group in KEDS. At the same time, errors involving the Konjunktiv II are greatly reduced in
Year C. This is mainly due to an improved grasp of the difference between the
morphology and meaning of indicative and Konjunktiv II modal verb forms, especially
KÖNNTE and WÜRDE, which is used to form the analytic Konjunktiv II (the most
common type of Konjunktiv II). Together with the underuse/overuse pattern between the
year groups, this indicates that the learners only start using the Konjunktiv II extensively,
once they feel confident that they have acquired the necessary morphological and
grammatical means. When this stage is reached, the learners seem keen to demonstrate
their abilities, which adds to the impression that learners have a propensity to explicitly
express modal meanings.
Another factor that makes the use of Konjunktiv II a useful indicator for learner
proficiency is the lexical development involved in this grammatical category. In the lower
proficiency groups (Years A and B) the use of Konjunktiv II is almost exclusively
restricted to modal verbs and the analytic WÜRDE-Konjunktiv (including the typical
errors with these). Konjunktiv II of auxiliary verbs (HABEN and SEIN) is rare, and lexical
verbs hardly occur at all in the Konjunktiv II. Despite being the smallest group, it is these
lexical verbs that are signifying growing proficiency most visibly. It has been discussed in
previous sections (see 6.2.1; 6.2.2; 6.2.4) that the use of certain modal items is more
indicative of spoken discourse than written. This can also be applied to Konjunktiv II
forms. In the theoretical outline of modal means (see 2.3.5.2). it was explained that the
use of synthetic Konjunktiv II forms in spoken language is usually restricted to a very
small set of very common verbs. These are kommen, gehen, wissen, geben, tun,
brauchen (come, go, know, give, do, need). In Year A, the only lexical verb in Konjunktiv
II form is geben, which occurs twice. In Year B, eight different lexical verbs are attempted
in the Konjunktiv II, but only five of these are successful (wissen, geben, kommen,
kotzen, träumen (know, give, come, puke, dream)). Apart from wissen, which occurs
twice, all of these only occur once in the corpus. We can see that, with only two
exceptions, all the lexical verbs that occur in the two lower proficiency groups are part of
the typical „spoken register‟ set. In Year C, learners expand their use of the Konjunktiv II
to fourteen different lexical verb types, which include several that are indicative of the
written register, e.g. liegen, heißen, verstehen, handeln (lie, mean, understand, act).
This, once more, shows a marked progression from the lower proficiency years (where a
writing style that is more akin to the spoken register has been noted also in other areas
265
of modality) to the more advanced group, where the learners show improved mastering
of grammatical structures as one factor of a more accomplished writing style.
There is, however, one area where there is not much progression to be observed in
the advanced year group, and that is the contexts in which the Konjunktiv II is used.
Whilst the overall frequencies of Konjunktiv II use increase dramatically towards Year C,
the proportional distribution in terms of context types still shows that learners
predominantly employ Konjunktiv II forms in the contexts that are typically used to teach
the Konjunktiv II, i.e. conditional clauses. In these contexts, they actually use Konjunktiv
II forms considerably more than the native speakers do. This also corresponds to a lower
percentage of use in general non-factual argumentation, i.e. not involved in conditional
contexts, which is lower than KEDS for all year groups.
In summary, learners in the lower years generally underuse Konjunktiv II and when
they do use it, it is mainly employed in „set pieces‟, i.e. conditionals. They use Konjunktiv
II less in general non-factual argumentation. There is some indication that, in these
contexts, the learners compensate with modal verbs (e.g. KÖNNEN) and periphrastic
modal expressions. In the more advanced group the learners feel more confident in
using the Konjunktiv II and consequently overuse it to explicitly express modality,
especially in conditional clauses.
This general tendency by all learners to overuse Konjunktiv II in conditionals could
either be materials induced, as the most salient form the learners encounter the
Konjunktiv II is in conditional clauses. Indeed, in the school text books that the learners
reported to have used in their A-level courses, the Konjunktiv II is actually called “the
conditional” (Brennpunkt) or even “the conditional tense” (Zeitgeist). It seems plausible,
then, that the learners associate the Konjunktiv strongly just with conditional contexts,
but not with any of the other possible contexts and also that they are confused about
Konjunktiv verb forms, as the textbooks use different terminology to refer to
morphological form (“subjunctive”) and function (“conditional”). This slightly inaccurate
linguistic description (the Konjunktiv is not just used in conditionals) might overall
contribute to the difficulties that learners exhibit with Konjunktiv verb forms, also on
modal verbs.
Another possibility for the high proportion of Konjunktiv II in conditional clauses is L1-
influence. Conditionals with explicit modality marking are more common in English than
266
in German, as the grammatical sequence of tense/mood and the indication of non-
factuality by means of a tense shift, is more prescriptive in English. In German, on the
other hand, hypothetical conditional clauses with indicative verb forms in both protasis
and apodosis are not just possible but also common, especially in spoken language. The
fact that these types of conditional clauses are underused by the learners, in conjunction
with the fact that errors with conditional clause patterns show influences of the L1
(learners have problems using Konjunktiv II in both apodosis and protasis in indirect
prospective or counter-factual conditional clauses) make a strong case for transfer here.
As it affects the way a certain grammatical structure is expressed in the syntax of the
target language, we could call the way learners employ conditional clauses an example
of syntactic transfer (see Jarvis and Pavlenko 2008:96f), although a point could also be
made for conceptual transfer as the „need‟ to express a hypothetical event as such (e.g.
by means of the Konjunktiv) could be interpreted as one that derives from the L1-
influenced conceptualisation of hypothetical events.
In summary, the analysis has shown that the Konjunktiv in German causes a
multitude of problems for learners that span across all areas of linguistic knowledge –
phonological, morphological, lexical, semantic and pragmatic. The very fact that it
challenges the learners on so many different levels marks it as a useful indicator of
proficiency. The pedagogical implications of these findings are not the subject of this
thesis, but Fabricius-Hansen (1997:32ff) provides detailed suggestions on how the
Konjunktiv could be taught in different stages to help learners cope with the multiple
demands that this structure poses.
6.3 CONCLUDING REMARKS
It should have become clear from the discussion in this chapter that there is no single
factor that influences the way learners utilise all the different modal means. In fact, even
within one type of modal means, e.g. modal verbs, there are several aspects that can
serve to explain the usage patterns of learners at different stages of proficiency.
Conversely, however, we have also found that many of these factors affect several types
of modal means in a similar way. This therefore suggest that they can be considered
267
general features of L2 argumentative writing, such as the tendency to be explicit and to
exhibit a more speech-like writing style than native speakers.
This study has mainly focused on the ways that the expression of modality in learner
language at different stages of proficiency differs from that of native speakers. One
concluding remark has to be added that addresses a fundamental similarity between L2
German writers and their native speaker counterparts. Several other studies have
investigated the cultural dependency of the use of epistemic modality in students‟ L2
writing. Hinkel (1995) and Gabrielatos and McEnery (2005) report that Asian students
generally exhibit a higher degree of assertiveness (e.g. through a reduced amount of
hedging devices; cf. Flowerdew 2000) in their argumentative style than English native
speakers and attribute this to the differences in the domains of sociocultural constructs
and presupposed values that these students operate within. They suggest that the
reasons for the divergence in argumentative style may lie in the different “culturally
bound understandings of the nature of obligation and necessity and adherence to
sociocultural norms and codes” (Hinkel 1995:337). McEnery and Kifle (2002), however,
found that Eritrean secondary school students used more weak epistemic modals and
tentatively attributed this to the influence of instructional materials.
In the present study, no such differences in the degree to which utterances are
asserted or mitigated could be detected between the learners and the native speakers.
On the one hand, this is not surprising, since the cultural background between the
English and German contributors to the corpora can be assumed to be similar with
respect to presupposed values and sociocultural norms and codes. On the other hand, it
might serve to refute the common conception that German native speakers are more
direct, brash and forceful in their argumentation than English native speakers.
This concludes the discussion of how the learner behaviour with regard to modal
expressions can be explained in terms of current second language acquisition theories.
In the following conclusion a general summary of the work carried out for this thesis is
presented. Then, an outlook on further investigations in this area is given as well as
some suggestions of how the findings from this study could be applied in the language
classroom in order to help learners improve their acquisition of modal expressions.
268
7. Conclusion
This study set out as an investigation of modality in learner language. A review of the
existing theoretical and empirical research on modality revealed that there is a
discrepancy between how modality is described in theoretical accounts and what is
investigated empirically. All different modal meanings can be expressed by a variety of
linguistic means; for the German language, these have been shown to include modal
verbs, a wide range of modal adverbials and lexical verbs, modal infinitives and the
Konjunktiv. The first aim set out in the introduction has therefore been achieved by
adopting a theoretical framework that allows for the inclusion of all types of modal
expressions.
The fact that modality can be expressed by more than just modal verbs is by no
means a new revelation, but for a long time much of the empirical research on modality
centred on modal verbs. It is only due to the advance of language-production focused
research, for instance within a corpus linguistic framework, that the importance and
interactions of all modal means have begun to be recognised and explored. It is, indeed,
the corpus linguistic approach that allows for the detection of the intricate usage patterns
of modal expressions in the first place. Only if all of these modal means are investigated
together, can we achieve a comprehensive picture of how the notion of modality is
conveyed in language.
This holds for native speaker language and learner language alike, so corpus
investigations can contribute new and insightful results to the study of second language
acquisition. Through frequency analyses we can, for example, establish whether one
type of modal expression is preferred over another in certain contexts. Investigations of
learner language have been shown to benefit greatly from this kind of approach,
especially if they are incorporated into a web of comparative analyses with native
speaker data of both the L1 and the L2, as well as other learner data. This is, essentially,
how the multiple-comparison approach applied in this study can provide insights into
learner language that other techniques cannot achieve. This substantiates the second
aim of the study, namely to demonstrate the usefulness and validity of a corpus
approach to the study of modality. This is not to say that other data gathering and
analysis procedures are less useful. Indeed, the kind of results that can be gleaned from
a learner corpus analysis cannot always provide conclusive proof of certain linguistic or
language acquisition theories. This is due to the fact that certain variables in the learner
269
data cannot be controlled for in return for the larger amount of data that can be managed
with corpus techniques as opposed to, for example, detailed qualitative analysis of
individual learners‟ output.
What it certainly can do, however, is contribute the kind of quantitative evidence that
other techniques cannot produce and therefore open up new angles of looking at and
comparing language learners‟ productions, which might corroborate or challenge findings
arrived at through different research designs. Frequency counts are just one factor in
this; learner corpus research also enables the investigator to defocus errors as the main
characteristic of learner language and put in the foreground the much richer picture of
both successful and unsuccessful learner performance.
In the present investigation, the result of this procedure is a detailed and intricate
account of how learners of German deal with modal expressions, which was set out as
the third aim of this research project. This uncovered specific patterns of overuse and
avoidance, which is one aspect that characterises how the learners‟ use of modality is
different from native speaker language. The analysis revealed, for example, that there is
a general overrepresentation of modal expressions in the learner data, that learners
greatly overuse the modal verb WERDEN and that they overuse certain periphrastic
modal expressions in epistemic meanings at the expense of modal verbs. Especially this
last point could only be detected, because the investigation included different types of
modal expressions and sought to investigate the nature of the interplay between them.
Furthermore, it could be shown on the basis of the contrastive analysis part of the
multiple-comparison approach that overuse and avoidance patterns of epistemic modal
verbs are strongly linked to L1/L2 form meaning equivalents, such as MUST and
MÜSSEN.
Another kind of pattern that gradually presented itself through the data-driven
approach to the learner data is learner specific formulaic sequences. Again, it is due to
the inclusion of as wide a range of modal expressions as possible that it could be
established that learners rely, for instance, heavily on sentence-initial Es ist... sequences
for conveying all kinds of different modal meanings. On the basis of these results it was
further suggested that learners generally do not have analytical meta-knowledge of the
category of modal infinitives, as they employ these almost exclusively within the
established Es ist... sequences.
The inclusion of material from different proficiency stages also allowed for statements
about how less and more advanced L2 writers employ modal expressions, from which
270
we can retrace developmental paths that the learners go through while advancing in
proficiency and pinpoint characteristics that distinguish advanced learners (e.g. the use
of the Konjunktiv). On the other hand, certain misuse patterns could be detected that
appear to be very resilient to progress even with the most advanced learner group, e.g.
the confusion between the modal verbs SOLLEN and SOLLTE.
On top of these findings that relate specifically to expressions of modality, evidence
could be gathered that underpins observations of general trends in learner language,
such as the tendency to be explicit and exhibit a less formal, more speech-like register in
the learners‟ writing. Taken together, all of these findings contribute to answering the first
two research questions of this study, namely how learners use modal expressions and
how this use differs at different stages of proficiency and from native speakers‟ use.
With regard to this last point of general features of learner language, the compilation
and analysis of a learner corpus for a language other than English has proven
particularly useful. The German learner corpus CLEG has provided evidence that can be
used to triangulate and thus corroborate findings of studies on L2 English that learners
exhibit certain general tendencies, for example the overuse of personalised expressions.
It could thus be established, that these trends are not just L1, but also L2-independent.
This is just one of the conclusions that could be drawn from the analysis in order to
answer the third research question on the possible reasons for the learners‟ linguistic
behaviour with regards to modal expressions. The discussion of the analysis results has
shown that many of them can be related to previous findings of learner language studies
as well as recent theories of second language acquisition. It has been shown that
several factors have to be considered as explanations for the learners‟ use of modal
expressions. L1-transfer, intra-lingual developmental factors and teaching materials all
play a role as they affect different types of modal expressions in different ways. On the
other hand, we have also seen that several of these factors can simultaneously affect the
same type of modal expressions. There are, however, also other aspects, such as
salience and morphological difficulties as well as general tendencies in learner language,
that influence the learners‟ specific usage of modality.
This study is certainly not without drawbacks and shortcomings. The fact that this
investigation was carried out within the confines of a PhD research project posed several
restrictions on the research design as well as the analysis. As the texts for the corpus
were collected and digitised by only one person, errors in the transcription of data are to
271
be expected despite every effort to eliminate these by spot checks. The limitations of a
single researcher affect the annotation of the data in the same way. Annotation errors
were corrected as much as possible as they became apparent during the detailed
analysis of the data, but ambiguous or doubtful cases had to be resolved without the
possibility of conferring with a team of researchers. This, on the one hand, almost
certainly leads to some disputable annotation decisions, but on the other hand it might at
least be considered internally consistent.
One of the main criticism that could be levelled against the design of the corpus itself
is that the division into proficiency groups is based on the external criteria of the year of
study alone. Not least from my own teaching experience am I very aware that these year
groups are by no means as homogenous in terms of proficiency as it might appear. This
is particularly true for the distinction between the first and second year of study. It would
therefore be desirable to group the students according to different proficiency criteria
when looking at developments from lower to higher proficiency groups. One of the ways
this could be done is by having the students complete a self-assessment questionnaire
based on the „can do‟-statements of the Common European Framework of Reference for
Languages (CEFR)155. This process had been partially carried out, but could not be
incorporated into the present study. Rather than having the students complete the
questionnaire, it might also be beneficial to have their tutors do this for them. Another
way would be to include marks for the individual texts or at least degree classes for each
learner. In my case, this was not possible due to confidentiality issues.
In terms of the texts that are included in the corpus, it has been pointed out before
that the second year group,Year B, constitutes a slightly different text profile to the other
two year groups as it contains some texts that are not exclusively argumentative. As
modality has been shown to be highly text-type sensitive, it had to be taken into
consideration that some of the results were influenced by this deviation in Year B. This
also has implications for the dispersion of modal expressions across the corpus, which in
turn reduces the indicativeness of the significance testing, as discussed in 4.3.3. As
significance testing was only used as additional evidence and not as the basis for the
inclusion or exclusion of items for investigation, this problem, while recognised, is not of
155
Cf. Council of Europe (2001). Common European Framework of Reference for Languages. Cambridge: Cambridge University Press; on the “can do”-statements see in particular: Jones, N and Hirtzel, M (2001) Appendix D: The ALTE „Can Do‟ Statements, in the Common European Framework of Reference for Languages: Learning, Teaching, Assessment. Council of Europe, Cambridge: Cambridge University Press.
272
major concern here. There are almost certainly more points of criticism that have not
been discussed here, but despite these shortcomings, many of which arise from practical
limitations, the design and execution of this study should have been rigorous enough to
produce valid and insightful results.
One of the main aims of this study was to widen the scope of the investigation to
modal expressions that had previously been neglected. While it can be ascertained that
the study incorporated a more comprehensive range of modal means than any other
study before, it is certainly not all-encompassing in every direction, which should provide
ample opportunity for further research.
It has to be acknowledeged that not every detail of every modal expression could be
examined in as much depth as it might deserve. Closer examinations within certain
modal meaning categories or certain groups of expressions might be worthwhile in order
to explore more thoroughly the pragmatic impact of the learners‟ use of modality. For
some of the findings it would also be fruitful to carry out investigations using different
methods of data gathering and analysis, e.g. elicitation tasks, in order to gain more
insights about the reasons why learners use certain modal expressions in the way they
do.
Another aspect that could not be discussed is, for example, how the different modal
means interact with each other in the same utterance. An investigation of
harmonic/disharmonic combinations in the learners‟ and native speakers‟ modal
expressions would undoubtedly contribute further to discovering learner specific traits
and features. For space reasons within the confines of this thesis, this issue could not be
pursued here, but would certainly provide a worthwhile follow-up investigation.
Furthermore, a factor that had to be neglected in the interpretation of the learner data
is “interlanguage transfer”, i.e. “the influence of one L2 (using the broad sense of this
term) over another” (Gass and Selinker 2008:152). For most of the contributors to the
CLEG corpus, German is not the first foreign language they are learning, but the second.
Most students started learning French before German. Additionally, many students take
up more than one language at university, mostly French and Spanish. As the design of
the learner corpus allows for factors like this to be taken into account, different
subcorpora could be compiled and compared according to other (previously acquired)
foreign languages. This could help answer questions with respect to possible differences
273
in the types of transfer that can be detected or to the application of interlanguage
knowledge of one foreign language in the use of another.
Finally, in addition to the two years‟ worth of learner data there is a further two years‟
available that has been collected during the writing of the thesis, but could not be
included in the present investigation. This means that, once this data is digitised, the
corpus contains truly longitudinal data from one complete cohort of students from their
first year through to their final year of studies. This should prove useful for more detailed
investigations into the writing development of individual learners. On top of this it could
help establish, through comparisons between the quasi-longitudinal and the truly
longitudinal data, whether extrapolations from quasi-longitudinal data regarding the
developmental paths of foreign language learners are justified.
To conclude, the presented study portrays a comprehensive picture of how learners of
German deal with expressions of modality as one aspect of language that is crucial to
the tasks they are required to perform in their university studies – argumentative writing.
It is beyond the scope of this thesis to draw pedagogical implications for the
improvement of learning, teaching and teaching materials. It is hoped, however, that the
results and discussions presented here may provide a useful basis for such enterprises.
Generally, a more direct and systemtic approach that incorporates more linguistic meta-
knowledge might be beneficial especially to British students, as they have often been
taught in a way that neglects the teaching of grammatical or semantic categories and
concepts. Rather than listing the modal verbs with their different meanings it might be
helpful to start with the semantic categories, e.g. possibility or necessity and then explore
the different means to express these. In this way, modal verbs can be introduced
alongside other modal means. In order to present the learners with a more accurate
representation of the distributions of different modal expressions, it might also be helpful
to use native speaker corpora in teaching, e.g. to discover the differences between
epistemic and non-epistemic uses of modal verbs and their semantic and grammatical
surroundings. With the periphrastic modal expressions, maybe one could even use the
learner corpus to show the differences in use and make evident to the learners the kinds
of formulaic patterns that they develop. This might encourage the students to refine their
use of modal expressions and encourage variability. Finally, much more attention should
be paid to phonological discrimination, again possibly in conjunction with more explicit
linguistic knowledge. Exercises that train the difference between umlaut and non-umlaut
274
vowels would help the students understand and handle the differences between
present, past and subjunctive modal verbs, but also with the whole category of the
German Konjunktiv, as the phonological and morphological distinctions go hand in hand
with actually understanding what the Konjunktiv does and how it is used.
275
Bibliography
Aarts, J. (1991). “Intuition-based and observation-based grammars”. In: Aijmer, K. and Altenberg, B. (eds). English Corpus Linguistics, Studies in Honour of Jan Svartvik. London: Longman. 44-62
Aarts, J. and Meijs, W. (1990)(eds). Theory and Practice in Corpus Linguistics. Amsterdam: Rodopi
Ackerman, J. (1991): “Reading, Writing, and Knowing: The role of Disciplinary Knowledge in Comprehension and Composing”. Research in the Teaching of English 25/2: 133-178
Admoni, V. (1982). Der deutsche Sprachbau. 4th revised and extended edition. Munich: Beck
Ahrenholz, B. (2000): “Modality and referential movement in instructional discourse. Comparing the production of Italian learners of German with native German and native Italian production”. In: Studies in Second Language Acquisition 22/3: 337-368
Aijmer, K. (2002). “Modality in advanced Swedish learners‟ written interlanguage”. In: Granger, S., Hung, J. and Petch-Tyson S. (eds). Computer Learner Corpora, Second Language Acquisition and Foreign Language Teaching. Amsterdam: Benjamins. 55-67
Alderson, C. (2007). “Judging the Frequency of English Words”. Applied Linguistics 28/3: 383-409
Altenberg, B. (1998). “Adverbial connectors in English and Swedish”. In: Hasselgård, H. and Oksefjell, S. (eds). Out of Corpora. Studies in Honour of Stig Johansson. Amsterdam: Rodopi. 249-268
Altenberg, B. and Tapper, M. (1998). “The use of adverbial connectors in advanced Swedish learners‟ written English”. In: Granger, S. (ed). Learner English on Computer. London – New York: Longman. 80-93
Augustin, I. (2006). Probleme mit deutschen Modalverben – Eine konstrastive Analyse zum Sprachgebrauch russisch sprechender Au-pair-Mädchen. Unpublished PhD thesis: Universität Bayreuth
Baker, P. (1997). “Consistency and Accuracy in Correcting Automatically Tagged Data”. In: Garside, R., Leech, G. and McEnery, T. (eds). Corpus Annotation. London – New York: Longman. 231-242
Bardovi-Harlig, K. (1987). “Markedness and Salience in Second-Language Acquisition”. Language Learning 37/3: 385-407
Bardovi-Harlig, K. (2000). Tense and aspect in second language acquisition: Form, meaning and use. Oxford: Blackwell
Barlow, M. (2002). “ParaConc: Concordance Software for Multilingual Parallel Corpora”. Retrieved June 20, 2009 from http://www.mt-archive.info/LREC-2002-Barlow.pdf
Basham, C. and Kwachka, P. (1991). “Reading the world differently: a cross cultural approach to writing assessment”. In: Hamp-Lyons, L. (ed). Assessing second language writing in academic contexts. Norwood, NJ: Ablex. 37-49
Bech, G. (1949). “Das semantische System der deutschen Modalverba”. Travaux du Cercle Linguistique de Copenhague 4: 3-46
Belz, J.A. (2004). “Learner Corpus Analysis and the Development of Foreign Language Proficiency”. System 32/4: 577-591
Biber, D. (2006). University language: A corpus-based study of spoken and written registers. Amsterdam: Benjamins
276
Biber, D. (1988). Variation across Speech and Writing. Cambridge: Cambridge University Press
Biber, D., Johansson, S., Leech, G., Conrad, S. and Finegan, E. (1999). Longman Grammar of Spoken and Written English. London: Longman
Biber, D., Conrad, S. and Reppen, R. (2004). Corpus Linguistics: Investigating Language Structure and Use. Cambridge: Cambridge University Press
Brinkmann, H. (1962). “Hochsprache und Mundart”. In: Wirkendes Wort, Sammelband 1. Düsseldorf: Schwann. 65-76
Brinkmann, H. (1971). Die Deutsche Sprache. Gestalt und Leistung. 2nd revised and extended edition. Düsseldorf: Schwann
Brünner, G. and Redder, A. (1983). Studien zur Verwendung der Modalverben. Tübingen: Narr
Bybee, J. and Fleischmann, S. (1995)(eds). Modality in Grammar and Discourse. Amsterdam – Philadelphia: Benjamins
Byrd, P. and Reid, J. (1998). Grammar in the Composition Classroom. Boston: Heinle & Heinle
Calbert, J. (1975). “Towards the Semantics of Modality”. In: Calbert, J. and Vater, H. Aspekte der Modalität. Tübingen: Narr. 1-70
Coates, J. (1983). The Semantics of Modal Auxiliaries. London – Canberra: Croom Helm Cumming, A. (1994)(ed). Bilingual Performance in Reading and Writing. Philadelphia:
Benjamins Dagneaux, E. (1995). Expressions of epistemic modality in native and non-native essay-
writing. Unpublished MA dissertation. Département d‟Etude Germaniques, Université Catholique de Louvain
Dagneaux, E., Denness, S. and Granger, S. (1998). “Computer-aided error anaysis”. System 26: 163-174
Dalmas, M. (1999). “Fakten und Effekte : Wozu gebraucht man eigentlich tatsächlich und Co.?” In: Wotjak, B. and Skibitzki, R. (eds.). Linguistik und Deutsch als Fremdsprache. Festschrift für Gerhard Helbig zum 70. Geburtstag. Tübingen: Niemeyer. 53-65
Dechert, H.W. (1983). “How a story is done in a second language”. In: Faerch, C. and Kasper, G. (eds). Strategies in interlanguage communication. New York: Longman.175-195
DeCock, S. (2000). “Repetitive phrasal chunkiness and advanced EFL speech and writing”. In: Mair, Ch. and Hundt, M. (eds). Corpus Linguistics and Linguistic Theory. Papers from the Twentieth International Conference on English Language Research on Computerized Corpora (ICAME 20) Freiburg im Breisgau 1999. Amsterdam – Atlanta: Rodopi. 51-68
DeCock, S., Granger, S., Leech, G. and McEnery, T. (1998). “An automated approach to the phrasicon of EFL learners”. In: Granger, S. (ed). Learner English on Computer. London – New York: Longman.67-79
Diewald, G. (1993). “Zur Grammatikalisierung der Modalverben im Deutschen”. Zeitschrift für Sprachwissenschaft 12/2: 218-234
Diewald, G. (1999). Die Modalverben im Deutschen. Grammatikalisierung und Polyfunktionalität. Tübingen: Niemeyer
Diewald, G. (2004). “Faktizität und Evidentialität: Semantische Differenzierungen bei den Modal- und Modalitätsverben im Deutschen”. In: Leirbukt, O. (ed). Tempus/Temporalität und Modus/Modalität im Sprachenvergleich. Tübingen: Stauffenburg. 231-258
277
Dittmar, N. (1979). “Fremdsprachenerwerb im sozialen Kontext. Das Erlernen von Modalverben – eine lexikalisch-semantische Analyse”. Sprache und Kontext. Zeitschrift für Literatur, Wissenschaft und Linguistik 33: 84-103
Dittmar, N. (1992). “Grammaticalization in second language acquisition: An introduction”. Studies in Second Language Acquisition 14: 249-257
Dittmar, N. (1993). “Proto-semantics and emergent grammars”. In: Dittmar, N. and Reich, A. (eds). Modality in language acquisition. Berlin: De Gruyter. 213-233
Dittmar, N. and Ahrenholz, B. (1995). “The acquistion of modal expressions and related grammatical means by an Italian learner of German in the course of 3 years of longitudinal observation”. In: Giacalone Ramat, A. and Crocco Galèas, G. (eds). From Pragmatics to Syntax. Modality in Language Acquisition. Tübingen: Narr. 197-232
Dittmar, N. and Reich, A. (eds)(1993). Modality in Language Acquisition. Berlin: De Gruyter
Dittmar, N. and Terborg, H. (1991). “Modality and second language learning: a challenge for linguistic theory”. In: Huebner, T and Ferguson, C.A. (eds). Crosscurrents in Second Language Acquisition and Linguistic Theories. Amsterdam: Benjamins. 347-384
Dodd, B. (2000)(ed). Working with German Corpora. Birmingham: Birmingham University Press
Doitchinov, S. (2001). ““Es kann sein, daß der Junge ins Haus gegangen ist” – Zum Erstspracherwerb von können in epistemischer Lesart”. In: Müller, R. and Reis, M. (eds). Modalität und Modalverben im Deutschen. Hamburg: Buske. 111-134
Dornseiff, F. (2004). Der Deutsche Wortschatz nach Sachgruppen. 8th revised edition. Berlin: DeGruyter
Doughty, C. and Williams, J. (eds)(1998). Focus on form in classroom second language acquisition. New York: Cambridge University Press
Dreyer, H. and Schmitt, R. (2000). Lehr- und Übungsbuch der deutschen Grammatik. München: Hueber
Droessinger, G. (2004). “Zum Begriff und zu den Arten der Modalität in der Linguistik”. Man and the Word (Žmogus ir žodis) 6/3: 85-94
Dunning, T. (1993). “Accurate Methods for the Statistics of Surprise and Coincidence”. Computational Linguistics 19-1: 61-74
Durrell, M. (2002). Hammer‟s German Grammar and Usage. Fourth revised edition. London: Arnold
Edmondson, W., House, J., Kasper, G. and McKeown, J. (1977). A Pedagogic Grammar of the English Verb. Tübingen: Narr
Ehlich, K. and Rehbein, J. (1972). “Einige Interrelationen von Modalverben”. In: Wunderlich, D. (ed). Linguistische Pragmatik. Frankfurt: Athenäum. 318-340
Ehrmann, M. (1966). The Meanings of Modals in Present-day American English. The Hague: Mouton
Ellis, N.C. (2008). “Usage-based and form-focused language acquisition”. In: Ellis, N.C. and Robinson P. (eds). Handbook of Cognitive Linguistics and Second Language Acquisition. New York – London: Routledge. 372-405
Engel, U. (1988). Deutsche Grammatik. 2nd amended edition. Heidelberg: Groos Erdmann, P. (1981). “Der Konjunktiv im britischen und amerikanischen Englisch”. In:
Kunsmann, P. and Kuhn, O. (eds.) Weltsprache Englisch in Forschung und Lehre. Festschrift für Kurt Wächtler. Berlin: Schmidt. 110-131
Fabricius-Hansen, C. (1997). “Der Konjunktiv als Problem des Deutschen als Fremdsprache”. In: Debus, F. and Leirbukt, O. (eds). Germanistische Linguistik.
278
Studien zu Deutsch als Fremdsprache: Aspekte der Modalität im Deutschen - auch in kontrastiver Sicht. Hildesheim: Olms. 13-36
Fagan, S. (2001). “Epistemic Modality and Tense in German”. Journal of Germanic Linguistics 13/3: 197-230
Fetzer, A. (2004). Recontextualising context: grammaticality meets appropriateness. Amsterdam – Philadelphia: Benjamins
Firth, J. R. (1956). “A new approach to grammar”. Reprinted in Palmer, F. R. (1968). Selected papers of J.R. Firth, 1952-1959. London: Oxford University Press. 206-209
Firth, J.R. (1951). “General Linguistics and Descriptive Grammar”. Transactions of the Philological Society 50/1: 69-71
Flowerdew, L. (2000). “Investigating referential and pragmatic errors in a learner corpus”. Burnard, L. and McEnery, T. (eds). Rethinking Language Pedagogy from a Corpus Perspective. Frankfurt: Lang. 145-154
Gabrielatos C. and McEnery T. (2005). “Epistemic modality in MA dissertations”. In: Fuertes Olivera P.A. (ed). Lengua y Sociedad: Investigaciones recientes en lingüística aplicada. Lingüística y Filología 61: 311-331
Gass, S. (1988). “Integrating research areas: a framework for second language studies”. Applied Linguistics 9: 198-217
Gass, S. and Selinker, L. (2008). Second Language Acquisition. An introductory course. Third edition. New York – London: Routledge
Gelhaus, H. (1977). Der modale Infinitiv. Mit einem dokumentarischen Anhang über die im gegenwärtigen Schriftdeutsch gebräuchlichen „bar“-Ableitungen. Tübingen: Narr
Geyken, A. (2009). “The DWDS corpus: a reference corpus for the German language of the 20th century”. In: Fellbaum, Ch. (ed). Idioms and Collocations. Corpus-based linguistic and lexicographic studies. London: Continuum. 23-40
Giacalone Ramat, A. (1992). “Grammaticalization process in the area of temporal and modal relations”. Studies in Second Language Acquisition 14: 297-322
Giacalone Ramat, A. (1995). “Function and form of modality in learner Italian”. In: Giacalone Ramat, A. and Crocco Galèas, G. (eds). From Pragmatics to Syntax. Modality in Language Acquisition. Tübingen: Narr. 269-293
Giacalone Ramat, A. (1999). “Grammaticalization of modality in language acquisition”. In: Studies in Language 23/2: 377-407
Giacalone Ramat, A. and Crocco Galèas, G. (eds)(1995). From Pragmatics to Syntax. Modality in Language Acquisition. Tübingen: Narr
Gilquin, G (2002). “Automatic retrieval of syntactic structures. The quest for the Holy Grail”. International Journal of Corpus Linguistics 7/2: 183-214
Gilquin, G. (2001). “The Integrated Contrastive Model. Spicing up your data”. Languages in Contrast 3(1): 95-123
Givón, T. (1979). On understanding grammar. New York: Academic Press Granger, S. (1994). “The learner corpus: a revolution in applied linguistics”. English
Today 39/10: 25-29 Granger, S. (1996). “From CA to CIA and Back: An Integrated Approach to
Computerized Bilingual and Learner Corpora”. In: Aijmer, K., Altenberg, B. and Johansson, M. (eds). Languages in Contrast. Papers from the a Symposium on Text-based Cross-linguistic Studies. Lund 4-5 March 1994. Lund: Lund University Press. 37-51
Granger, S. (ed)(1998a). Learner English on Computer. London – New York: Longman Granger, S. (1998b). “Prefabricated patterns in advanced EFL writing: Collocations and
formulae”. In: Cowie, A.P. (ed). Phraseology: Theory, analysis and applications. Oxford: Oxford University Press. 145-160
279
Granger, S. (2002). “A Bird‟s eye view of learner corpus research”. In: Granger, S., Hung, J. and Petch-Tyson, S. (eds). Computer Learner Corpora, Second Language Acquisition and Foreign Language Teaching. Amsterdam: Benjamins. 3-33
Granger, S., Hung, J. and Petch-Tyson, S. (eds)(2002). Computer Learner Corpora, Second Language Acquisition and Foreign Language Teaching. Amsterdam: Benjamins
Granger, S. and Rayson, P. (1998). “Automatic profiling of learner texts”. In: Granger, S. (ed). Learner English on Computer. London – New York: Longman. 119-131
Gutknecht, C. and Rölle, L.J. (1988). “Die multifaktorielle Translationssituation bei den Modalverben des Sprachenpaares Deutsch-Englisch”. In: Quast, G. (ed). Einheit in der Vielfalt. Festschrift für Peter Lang zum 60. Geburtstag. Bern – Frankfurt/M. – New York – Paris: Lang. 154-215
Halliday, M.A.K. (1970). “Functional diversity in language as seen from a consideration of modality and mood in English”. Foundations of Language 6: 322-365
Halliday, M.A.K. (1992). “Language as system and language as instance: the corpus as a theoretical construct”. In: Svartvik, J. (ed). Directions in Corpus Linguistics. Berlin: deGruyter. 61-77
Han, Z-H. (2000). “Persistence of the implicit influence of NL: The case of the pseudo-passive”. Applied Linguistics 21/1: 78-105
Hasselgren, A. (1994). “Lexical Teddy Bears and Advanced Learners: a study into the ways Norwegian students cope with vocabulary”. International Journal of Applied Linguistics 4: 237-260
Heine, B. (1995). “Agent-oriented vs. Epistemic Modality. Some observations on German Modals”. In: Bybee, J. and Fleischmann, S. (eds). Modality in Grammar and Discourse. Amsterdam – Philadelphia: Benjamins. 17-53
Heine, B., Claudi, U. and Hünnemeyer, F. (1991). Grammaticalization: A Conceptual Framework. Chicago: University of Chicago Press
Heinrichs, W. (1981): Die Modalpartikeln im Deutschen und Schwedischen. Eine Linguistische Analyse. Linguistische Arbeiten, Band 101. Tübingen: Niemeyer
Helbig, G. (2007). “Der Konjunktiv und kein Ende. Zu einigen Kontroversen in der Beschreibung des Konjunktivs der deutschen Gegenwartssprache”. Deutsch als Fremdsprache 3: 140–153
Helbig, G. and Buscha, J. (2001). Deutsche Grammatik. Ein Handbuch für den Ausländerunterricht. Berlin – München: Langenscheid
Helbig, G. and Helbig, A. (1990). Lexikon deutscher Modalwörter. Leipzig: Verlag Enzyklopädie
Hermann, Ch., McCrorie, M. and Sauer, D. (2001). Zeitgeist. Oxford: Oxford University Press
Hinkel, E. (1995). “The Use of Modal Verbs as a Reflection of Cultural Values”. TESOL Quaterly 29/2: 325-343
Hinkel, E. (2009). “The effect of essay topics on modal verb uses in L1 and L2 writing. Journal of Pragmatics 41: 667-683
Hofmann, T.R. (1979). “On Modality in English and other Languages”. Papers in Linguistics 12: 1-37
Holl, D. (2001). “Was ist modal an Modalen Infinitiven?” In: Müller, R. and Reis, M. (eds). Modalität und Modalverben im Deutschen. Hamburg: Buske. 217-238
Holmes, J. (1988). “Doubt and certainty in ESL textbooks”. Applied Linguistics 9: 21-44 Hoye, L. (1997). Adverbs and Modality in English. London – New York: Longman Hunston, S. (2002). Corpora in Applied Linguistics. Cambridge: Cambridge University
Press
280
Hyland, K. and Milton, J. (1997). “Qualification and certainty in L1 and L2 students‟ writing”. Journal of Second Language Writing 6/2: 183-205
Itayama, M. (1993). “Werden: Modaler als die Modalverben!”. Deutsch als Fremdsprache 30: 233-37
James, C. (1998): Errors in Language Learning and Use. Exploring Error Analysis. London-New York: Longman
Jarvis, R.A. (1972): “A pedagogical grammar of the modal auxiliaries”. English Language Teaching 26: 238-248
Jarvis, S. (2000). “Methodological rigor in the study of transfer: Identifying L1 influence in the interlanguage lexicon”. Language Learning 50: 245-309
Jarvis, S. and Pavlenko, A. (2008). Cross-linguistic influence in language and cognition. New York – London: Routledge
Jarvis. S. and Odlin, T. (2000). “Morphological type, spatial reference, and language transfer”. Studies in Second Language Acquisition 22: 535-556
Jiang, N. (2002): “Form-meaning mapping in vocabulary acquisition in a second language”. Studies in Second Language Acquisition 24: 617-637
Kärkkäinen, E. (1992). “Modality as a strategy in interaction: Epistemic modality in the language of native and non-native speakers of English”. In: Bouton, L.F. and Kachru, Y. (eds). Pragmatics and language learning, volume 3. Urbana, IL: Urbana-Champaign. 197-216
Kasper, G. and Rose, K. (2002). Pragmatic Development in a Second Language. Oxford: Blackwell
Kątny, A. (1993). “Lexical and Grammatical Exponents of Modality in Polish and German”. In: Dittmar, N. and Reich, A. (eds). Modality in Language Acquisition. Berlin: De Gruyter. 41-57
Kratzer, A. (1978). Semantik der Rede. Kontexttheorie, Modalwörter, Konditionalsätze. Königstein: Scriptor.
Kratzer, A. (1991). “Modality”. In: Stechow, A.v. and Wunderlich, D. (eds) Semantik. Ein internationales Handbuch der zeitgenössischen Forschung. Berlin – New York: De Gruyter. 639-650
Kratzer, A. (forthcoming). Modals and Conditionals Again. Oxford: Oxford University Press
Kufner, H. (1962). The Grammatical Structures of English and German. Chicago: University of Chicago Press
Kwachka, P. and Besham, C. (1990). “Literacy acts and cultural artifacts: on extensions of English modals”. Journal of Pragmatics 14: 413-429
Lakoff, G. (1987). Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago: University of Chicago Press.
Lambert, M. (1995). “Variabilité dans le recours aux formes personnelles et impersonnelles dans l‟expression de la modalité épistémique par des apprenants avancés en anglais”. In: Giacalone Ramat, A. and Crocco Galèas, G. (eds). From Pragmatics to Syntax. Modality in Language Acquisition. Tübingen: Narr. 233-248
Lambrecht, K. (1994). Information structure and sentence form: topic, focus, and the mental representations of discourse referents. Cambridge: Cambridge University Press
Langhoff, S. (1980). Gestaltlinguistik. Eine ganzheitliche Beschreibung syntaktisch-semantischer Sprachfunktionen am Beispiel Modaler Infinitivkonstruktionen des Deutschen und Englischen. Frankfurt: Lang
Lauridsen, O. and Poulsen, S. (1995). Tysk Grammatik. Kopenhagen:Munksgaard Leech , G. (1971). Meaning and the English Verb. London: Longman
281
Leech, G (1992). “Corpora and theories of linguistic performance”. In: Savrtvik, J. (ed). Directions in Corpus Linguistics. Berlin: deGruyter. 125-148
Leech, G. (2003). “Modality on the move: the English modal auxiliaries 1961-1992”. In: Facchinetti, R., Krug, M. and Palmer, F.R. (eds). Modality in contemporary English. Berlin – New York: Mouton de Gruyter
Leech, G. and Svartvik, J. (1975). A Communicative grammar of English. London: Longman
Lemnitzer, L. and Zinsmeister, H. (2006). Korpuslinguistik: Eine Einführung. Tübingen: Narr
Lindemann, B. (1996). Der Erwerb des deutschen Modalversystems: Systematizität und Variationen bei der Entwicklung der individuellen Lernersprachen norwegischer Deutschlerner. Tübingen: Narr
Lorenz, G. (1998). “Overstatement in advanced learners‟ writing: stylistic aspects of adjective intensification”. In: Granger, S. (ed). Learner English on Computer. London – New York: Longman. 53-66
Lorenz, G. (1999). Adjective intensification – Learners versus Native Speakers. A corpus study of argumentative writing. Amsterdam – Atlanta: Rodopi
Lüdeling, A. (2008). “Mehrdeutigkeit und Kategorisierung: Probleme bei der Annotation von Lernerkorpora”. In: Walter, M. and Grommes, P. (eds). Fortgeschrittene Lernervarietäten. Korpuslinguistik und Zweitspracherwerbsforschung. Tübingen: Niemeyer. 119-140
Lüdeling, A., Doolittle, S., Hirschmann, H., Schmidt, K. and Walter, M. (2008). “Das Lernerkorpus Falko”. Deutsch als Fremdsprache 2/2008: 67-73.
Lüdeling, A., Walter, M., Kroymann, E. and Adolphs, P. (2005). “Multi-level error annotation in learner corpora”. In: Proceedings of Corpus Linguistics 2005 (Birmingham)
Lyne, A. (1985). The vocabulary of French business correspondence: word frequencies, collocations, and problems of lexicometric method. Paris: Champion
Lyons, J. (1977). Semantics. Cambridge: Cambridge University Press MacWhinney, B. (2000). The CHILDES Project: Tools for Analyzing Talk.
Hillsdale, NJ: Erlbaum McCrostie, J. (2008). Writer visibility in EFL learner academic writing: A corpus based
study. ICAME Journal 32: 97-114 McEnery, T. and Kifle, N.A. (2002). “Epistemic modality in argumentative essays of
second-language writers”. In: Flowerdew, J. (ed). Academic Discourse. Harlow: Longman. 182-215
McEnery, T. and Wilson, A. (1996). Corpus Linguistics. Edinburgh: Edinburgh University Press
Merkel, M. (2001). Comparing source and target texts in a translation corpus. Presented at the 13th Nordic Conference on Computational Linguistics, NoDaLiDa'01, Uppsala, Sweden
Meunier, F. (2000). A computer corpus linguistics approach to interlanguage grammar: noun phrase complexity in advanced learner writing. Unpublished PhD thesis. Centre for English Corpus Linguistics. Université catholique de Louvain, Louvain-la-Neuve
Mindt, D. (1996). “English corpus linguistics and te foreign language teaching syllabus”. In: Thomas, J. and Short, M. (eds). Using corpora for language research. London – New York: Longman. 232-247
Mitchell, R. and Martin, C. (1997). “Rote learning, creativity and „understanding‟ in classroom foreign language teaching”. Language Teaching Research 1/1: 1-27
282
Mortelmans, T. (1997). “Konjunktiv II and Epistemic Modals in German. A Division of Labour”. In: Foolen, A. and van der Leek, F. (eds). Constructions in Cognitive Linguistics. Amsterdam – Philadelphia: Benjamins. 191-216
Myles, F. (2005). “Interlanguage corpora and second language acquisition research”. Second Language Acquisition Research 21/4: 373-391
Nakamura, J. (1993). “Statistical methods and large corpora: A new tool for describing text types”. In Baker, M., Francis, G. and Tognini-Bonelli, E. (eds). Text and Technology. Amsterdam: Benjamins. 313-332
Narrog, H. (2005). “On Defining Modality Again”. In: Language Sciences, 27 (2): 165-192 Neff van Aertselaer, J. (2008). “Contrastive English-Spanish interpersonal discourse
phrases: A corpus study”. In: Granger, S. and Meunier, F. (eds). Phraseology in foreign language learning and teaching. Amsterdam: Benjamins. 85-99
Nehls, D. (1986). Semantik und Syntax des englischen Verbs. Teil II: Die Modalverben. Eine kontrastive Analyse der Modalverben im Englischen und Deutschen. Heidelberg: Groos
Nesselhauf, N. (2004). “Learner corpora and their potential for language teaching”. In: Sinclair, J. (ed). How to Use Corpora in Language Teaching. Amsterdam: Benjamins. 125-152
Odlin, T. (2005). “Cross-linguistic influence and conceptual transfer: What are the concepts?” Annual Review of Applied Linguistics 25: 3-25
Öhlschläger, G (1984). “Modalität im Deutschen”. Zeitschrift für Germanistische Linguistik 12/2: 229-246
Palmer, F.R. (1979). Modality and the English Modals. London: Longman Palmer, F.R. (1986). Mood and Modality. Cambridge: Cambridge University Press Pavlenko, A. and Diragina, V. (2007). “Russian emotion vocabulary in American learners'
narratives”. Modern Language Journal 91/2: 213-234 Perkins, M.R. (1983). Modal Expressions in English. London: Pinter Petch-Tyson, S. (1998). “Writer/reader visibility in EFL written discourse”. In: Granger, S.
(ed). Learner English on Computer. London – New York: Longman. 107-118 Pfeffer, J.A. and Conermann, J. (1977). “Das Futur mit und ohne werden in Wort und
Schrift”. Die Unterrichtspraxis 10: 78-87 Pravec, N. A. (2002). “Survey of learner corpora”. ICAME Journal 26: 81-114 Quirk, R., Greenbaum, S., Leech, G. and Svartvik, J. (1985). A Comprehensive
Grammar of the English Language. London: Longman Raimes, A. (1994).“Language Proficiency, Writing Ability, and Composing Strategies: A
Study of ESL College Student Writers”. In: Cumming, A. (ed). Bilingual Performance in Reading and Writing. Philadelphia: Benjamins. 139-172
Rayson P., Berridge D. and Francis B. (2004). “Extending the Cochran rule for the comparison of word frequencies between corpora”. In: Purnelle G., Fairon C., Dister A. (eds.) Le poids des mots: Proceedings of the 7th International Conference on Statistical analysis of textual data (JADT 2004). 10-12 March 2004, Louvain-la-Neuve, Belgium. Presses universitaires de Louvain. 926-936
Rayson, P. and Garside, R. (2000). “Comparing corpora using frequency profiling”. In: Proceedings of the workshop on Comparing Corpora, held in conjunction with the 38th annual meeting of the Association for Computational Linguistics (ACL 2000). 1-8 October 2000, Hong Kong.
Redder, A. (2001). “Modalverben in wissenschaftlicher Argumentation: Deutsch und Englisch im Vergleich”. Jahrbuch Deutsch als Fremdsprache 27: 313-330
Rescher, N. (1968). Topics in philosophical logic. Dordrecht: Reidel Ringbom, H. (1987). The Role of the First Language in Foreign Language Learning.
Clevedon: Multilingual Matters
283
Ringbom, H. (1998). “Vocabulary frequencies in advanced learner English: a cross-linguistic approach”. In: Granger, S. (ed). Learner English on Computer. London – New York: Longman. 41-52
Salkie, R. (1995). “INTERSECT: a parallel corpus project at Brighton University”. Computers & Texts 9: 4-5
Salkie, R. (1997). “Naturalness and contrastive linguistics”. In: Lewandowska-Tomaszczyk, B and Melia, P. J. (eds). Proceedings of PALC '97. University of Lodz 1997. 297-312
Salsbury, T. and Bardovi-Harlig, K. (2000). “Oppositional talk and the acquisition of modality in L2 English”. In: Swierzbin, F., Morris M., Anderson, M., Klee, C.A. and Tarone, E. (eds). Social and cognitive factors in second language acquisition. Somerville, MA: Cascadilla Press. 57-76
Sapir, E. (1958): Culture, Language and Personality (ed. D. G. Mandelbaum). Berkeley, CA: University of California Press
Schachter, J. and Rutherford, W. (1979). “Discourse function and language transfer”. Working Papers in Bilingualism 19: 3-12
Schmid, H. (1997). “Probabilistic part-of-speech tagging using decision trees”. In: Jones, D.B. and Somers, H.L. (eds). New Methods in Language Processing. New York – London: Routledge, 154-164
Schmid, H.-J. (2001). “„Presupposition can be a bluff‟: How abstract nouns can be used as presupposition triggers”. Journal of Pragmatics 33/10: 1529-1552
Schmidt, R. (1990). “The role of consciousness in second language learning”. Applied Linguistics 11: 129-158
Schmidt, R. (1995). “Consciousness and foreign language learning: A tutorial on the role of attention and awareness in learning.” In: Schmidt, R. (ed). Attention and awareness in foreign language learning. Honolulu: University of Hawai‟i, Second Language Teaching & Curriculum Center. 1-63
Schmidt, R. (2001). “Attention”. In: Robinson, P. (ed). Cognition and second language instruction. New York: Cambridge University Press. 3-32
Schmied, J. and Schäffler, H. (1996). “Approaching translationese through parallel and translation corpora”. In: Percy, E., Meyer, C.F. and Lancashire, I. (eds). Synchronous corpus linguistics. Papers from the Sixteenth International Conference on English Research on Computerized Corpora (ICAME 16). Amsterdam: Rodopi, 41-58
Schütze, C.T. (1996). The empirical base of linguistics: grammaticality judgments and linguistic methodology. Chicago: University of Chicago Press
Schulz, D. and Griesbach, H. (1955). Deutsche Sprachlehre für Ausländer. Munich: Hueber
Scott, M. (2008). WordSmith Tools version 5. Liverpool: Lexical Analysis Software Scott, M. and Tribble, C. (2006). Textual Patterns: Key Words and Corpus Analysis in
Language Education. Philadelphia: Benjamins Selinker, L. (1972): “Interlanguage”. International Review of Applied Linguistics 10/3:
209-231 Sharwood Smith, M. (1981). “Consciousness raising and the second-language learner”.
Applied Linguistics 2: 159-168 Sheperd, S. (1993). “The acquisition of modality in Antiguan Creole.” In: Dittmar, N. and
Reich, A. (eds). Modality in Language Acquisition. Berlin: De Gruyter. 171-184 Sinclair, J. (1991). Corpus, concordance, collocation. Oxford: Oxford University Press Skiba, R. (1993). “Modal Verbs and Their Syntactic Characteristics in Elementary
Learner Varieties”. In: Dittmar, N. and Reich, A. (eds.). Modality in language acquisition. Berlin: De Gruyter. 247-259
284
Skiba, R. and Dittmar, N. (1992). “Pragmatic, semantic, and syntactic constraints and grammaticalization: A longitudinal perspective.” Studies in Second Language Acquisition 14: 323-349
Smith, K. (2008). “Teaching Talented Writers in the Regular Classroom”. Gifted Child Today 31/ 2: 19-26
Stechow, A.v. and Wunderlich, D. (eds)(1991). Semantik. Ein internationales Handbuch der zeitgenössischen Forschung. Berlin – New York: De Gruyter
Steele, S., Akmajian, A., Demers, R., Jelinek, E., Kitagawa, C., Oerle, R. and Wasow, T. (1981). An encyclopedia of AUX: a study in cross-linguistic equivalence. Cambridge, MA – London: MIT Press
Stephany, U. (1995). “Function and form of modality in first and second language acquisition”. In: Giacalone Ramat, A. and Crocco Galèas, G. (eds)(1995). From Pragmatics to Syntax. Modality in Language Acquisition. Tübingen: Narr. 105-120
Stevik, E.W. (1989). Success with foreign languages. Hemel Hempstead: Prentice Hall Stubbs, M. (1993). “British traditions in text analysis: from Frith to Sinclair”. In: Baker, M.,
Francis, G. and Tognini-Bonelli, E. (eds). Text and Technology. In Honour of John Sinclair. Amsterdam: Benjamins. 1-33
Sundry, C., Somerville, J., Morris, P. and Aberdeen, H. (2000). Brennpunkt. Cheltenham: Nelson Thornes
Svartvik, J. (1992). Directions in Corpus Linguistics. Berlin: deGruyter Sweetser, E. E. (1982). “Root and epistemic modals: causality in two worlds”. Berkeley
Linguistic Society Papers 8: 484-507 Sweetser, E. E. (1990). From etymology to pragmatics: Metaphorical and cultural
aspects of semantic structure. Cambridge: Cambridge University Press Takahashi, K. (2006). “A Study of Register Variation in the British National Corpus”.
Literary and Linguistic Computing 21/1: 111-126 Taylor, G. (1986). “Errors and explanations”. Applied Linguistics 7/2: 144-166 Terborg, H. (1993). “Elaboration of the Modal Lexicon in German as a Second
Language”. In: Dittmar, N. and Reich, A. (eds). Modality in language acquisition. Berlin: De Gruyter. 235-245
Terrell, T. (1991). “The role of grammar instruction in a communicative approach”. The Modern Language Journal 75: 52-63
Tono, Y. (2002): The Role of Learner Corpora in SLA Research and Foreign Language Teaching: A Multiple Comparison Approach. Unpublished PhD Thesis, Lancaster University
Tono, Y. (2002a). Learner Corpus Research: State of the Art and Applications in SLA Research. Invited talk given at the Japanese Society of Language Sciences at Nihon Women's University. Retrieved May 13, 2009 from http://leo.meikai.ac.jp/~tono/
Traugott, E. and Heine, B. (eds)(1991.1, 1992.2). Approaches to Grammaticalization. 2 volumes. Amsterdam – Philadelphia: Benjamins
Traugott, E. and König, E. (1991). “The Semantic-Pragmatics of Grammaticalization Revisited.” In: Traugott, E. and Heine, B. (eds). Approaches to Grammaticalization. Amsterdam – Philadelphia: Benjamins. 189-218
Vethamani, M., Manaf, U. and Akbari, O. (2008). “Students‟ Use of Modals in Narrative Compositions: Forms and Functions”. In: English Language Teaching 1/1: 61-74
Viana, V. (2006). “Modals in Brazilian Advanced EFL Learners‟ Compositions: A Corpus-Based Investigation”. Profile 7: 77-86
von Wright, G.H. (1951). An essay in modal logic. Amsterdam: North Holland Walter, M. and Grommes, P. (2008). “Die Entdeckung des fortgeschrittenen Lerners in
der Varietätenlinguistik”. In: Walter, M. and Grommes, P. (eds). Fortgeschrittene
285
Lernervarietäten. Korpuslinguistik und Zweitspracherwerbsforschung. Tübingen: Niemeyer. 3-28
Wehrle, H. and Eggers, H. (1993). PONS Deutscher Wortschatz. Ein Wegweiser zum treffenden Ausruck. Stuttgart: Klett
Weinberger, U. (2002). Error Analysis with Computer Learner Corpora: A corpus-based study of errors in the written German of British university students. Unpublished Master‟s Thesis. University of Lancaster
Weinberger, U. (2005). “Error analysis with computer learner corpora: a corpus-based study of errors in the written German of British university students”. In: Allerton, D.J., Tschichold, C. and Wieser, J. (eds). Linguistics, Language Learning, Language Teaching. Basel: Schwabe. 119-130
Weinrich, H. (2005). Textgrammatik der deutschen Sprache. 3rd revised edition. Hildesheim – Zürich – New York: Olms
Welke, K. (1965). Untersuchungen zum System der Modalverben in der deutschen Sprache der Gegenwart. Ein Beitrag zur Erforschung funktionaler und syntaktischer Beziehungen. Berlin: Akademie
Werlich, E. (1976). A text grammar of English. Heidelberg: Quelle & Meyer Whorf, B.L. (1956): Language, Thought and Reality (ed. J. B. Carroll). Cambridge, MA:
MIT Press Wichmann, A. and Nielsen, J. (2000). “Rights and obligations in legal contracts: corpus
evidence”. In: Dodd, B. (ed). Working with German Corpora. Birmingham: Birmingham University Press. 245-266
Wilson, A. (forthcoming). “Speech, Writing and Discourse Type”. In: Culpeper, J., Katamba, F., Kerswill, P., Wodak, R. and McEnery, T. (eds). English Language. Description, Variation and Context. Basingstoke: Palgrave Macmillan
Witton, N. (2000). “Inflected and periphrastic subjunctive verb forms in German newspaper texts of the 1960s and 1990s”. In: Dodd, B. (ed). Working with German Corpora. Birmingham: Birmingham University Press. 267-298
Wray, A. (2002). Formulaic Sequences and the Lexicon. Cambridge: Cambridge University Press
Xiaoling, J. and Xiao, L. (2008). “A comparative study of modal verb use between EFL and ENL essays”. Foreign Language Education in China 1/3: 18-26
Zifonun, G., Hoffman, L. and Strecker, B. (1997). Grammatik der deutschen Sprache. Schriften des Instituts für deutsche Sprache Bd. 7. Berlin – New York: De Gruyter
286
Appendix
APPENDIX A
EXAMPLES OF LEARNER TEXTS:
EXAMPLE OF A “KRITISCHE ZUSAMMENFASSUNG” (CRITICAL SUMMARY)
[a1023_04]
Kinder und Computers - es hängt von der Dosis
Obwohl fast alle deutsche Haushalten haben mindestens ein Computer, es könnte
beides Positive und Negative Folgen haben, besonders für Kinder. Sie können nicht nur
intelligenz fordern sondern auch zu steigende Bewegungsarmut führen.
Es gibt viele Lernprogramme für junge Kinder aber es gibt auch viele Sorge von
Kinderärzte, die diese ungesunde Kinder sehen müssen. Die Kinder sind immer mehr
übergewichtig, weil sie Stubenhocker sind. Aber die Meisten glauben, dass Computers
sehr gut für die Entwicklung dieser Kinder sind, wenn sie nicht zu viel zeit vor das
Computer verbringen.
Manche Leute sagen auch, dass eben nur einfachen Spiele Intelligenz fordern können.
Kinder in die Grundschule, die Computers benutzen können besser rechnen und haben
besser Feinmotorik, zeichnen wissenschaftliche Studien. Währscheinlich sind das
Fernsehen das Problem, nicht Computers.
Auf eine Seite möchten besondere junge Kinder etwas Aktiv zu machen nach ein paar
Minuten. Dann fahren sie rad zum Beispiel. Aber auf die andere Seite gibt es viele
Sorgen, dass Kinder nur in ihre Haus setzen, deshalb spielen sie nicht. Diesen können
zu eine schlechte Sprachenentwicklung führen, weil sie immer allein setzen. Es gibt
auch keine Familienessen mehr. Aber viele akzeptieren, dass Computers können, zum
Beispiel, die Fingerfertigkeit verbessern. Nochmal kann man etwas gegen diese
Meinungen sagen. Das Computer ist sehr nützlich für Kinder, weil ihre Freunden können
auch auf das Computer spielen und so weite. Jetzt mögen immer mehr Mädchen
Computers auch. Es könnte sein, dass das Problem hauptlich mit Jugendlichen ist, nicht
mit junge Kindern, obwohl sie nur für 45 minuten bis eine Stunde pro Tag vor dem
Computer setzen sollen. Die Schule in Deutschland betutzen Computer kaum in die
Unterrichten. Viele leute sagen, dass es würde sehr nützlich sein, Computers in die
Schule zu benutzen, besonders für Kinder mit Lernprobleme und der Legasthenie.
287
EXAMPLE OF A CRITICAL COMMENTARY (“KRITISCHER KOMMENTAR”)
[c2044_01]
Sollte der Einsatz von Gentechnik gestoppt werden?
Gentechnik ist seit Jahren ein kontroverses Thema. In vielen Ländern versuchen jetzt
Supermärkte weniger genetisch veränderte Nahrungsmitteln in ihren eigenen Produkten
zu verwenden, aber Gentechnik bedeutet nicht nur die Veränderung von
Nahrungsmitteln, sondern auch die Veränderung von Organismen und das Erbgut der
Menschen. Der Artikel berichtet über die Weiterentwicklungen der Gentechnik und die
Konsequenzen für das deutsches Volk.
Es ist nicht zu leugnen, dass der Einsatz von Gentechnik positive Wirkungen für
Medikamenten hat. Der Umstand, dass schon über 40 Medikamente aus gentechnischer
Produktion erhältlich sind, die nie früher hergestellt werden konnten, kann sicherlich nur
eine gute Sache sein. Genau so wichtig ist, dass wir jetzt verstehen wie die Gene
wirken, sowohol in Erb- und Volkskrankheiten wie in Viruserkrankungen wie Aids, das
allein letztes Jahr 3.1 Million Menschen weltweit getötet hat'. Deshalb sind mit Hilfe von
Gentechnik Heilmittel für solche, früher unheilbare, Erkrangungen sind jetzt „im Visier".
Hinzu kommt, dass es nicht nur Heilungschancen sind, sondern auch offensichtlich
diagnostischen Möglichkeiten. Wenn eine Krankheit früher diagnostiziert werden könnte,
gibt es für den Patient weniger Belastung. Sie können sich für Heilbehandlung besser
vorbereiten, und der ganze Prozess dauert dann hoffentlich nicht so lang, da der Virus
noch nicht in einem so weit fortgeschrittenen Stadium sein wird. Aber wenn wir die
diagnostischen Möglichkeiten genauer betrachten, sehen wir wohl auch Nachteile. Es
hat mit der Vertrauenswürdigkeit von Leuten zu tun, wie zum Beispiel Ärzten oder
Pharma-Firmen.
Trotz dieser Argumente, die wohl unter anderem für die Einführung der Gentechnik
ausschlaggebend waren, müssen wir auch die Risiken von Gentechnik hervorheben.
Felder mit gentechnisch veränderten Pflanzen werden immer noch zerstört. Warum
machen die Leute so etwas? Vielleicht passiert es, weil die Menschen, die in der Nähe
dieser veränderten Pflanze leben, kein Risiko für ihre Gesundheit eingehen wollen. Wir
wissen auch nicht viel über die Nebenwirkungen der Gentechnik, wenn wir zum Beispiel
Medikamente nehmen oder genetisch veränderte Lebensmittel essen. Es ist schwierig
die Langzeitwirkung zu untersuchen, da die Technik selbst ziemlich neu ist.
Deutschland hat allerdings Probleme mit gentechnishen Versuchen. Sie sind immer
noch mit der Nazi-Partei verbunden, besonders wenn es um den Bereich der
288
Verbesserung oder Änderung menschlichen Lebens geht. Viele Leute sind ums Leben
gekommen in Konzentrationslagern wie Dachau, als Folge der Behandlung durch Nazi-
Ärzte , die versuchten den "perfekten" Deutschen zu erschaffen. Versuchen sind aber
nötig, bevor wir Gentechnik Prozesse, die sich mit dem Erbgut des Menschens
beschäftigen, durchführen können. Solche Prozesse müssen mit der äußerste Präzision
durchgeführt werden und Versuche sollten langfristig gemacht werden, damit wir
Langzeitwirkung identifizieren können.
Wir müssen aber auch beachten, dass wenn den Umgang mit menschlichen Embryonen
für Versuche erforderlich ist, es mit sich nicht nur erhebliche Risiken bringt, sondern es
auch moralische Fragen aufwirft.
Letztendlich stellt sich die Frage: Wie weit dürfen wir gehen in der Gentechnologie? Ist
es richtig und zulässig, bestimmte Krankheiten mit Hilfe der Gentechnik zu bekommen?
Die meisten Leuten werden zustimmen. Ist es aber auch richtig, Menschen nach
Wünsch zu züchten, zum Beispiel das gesunde Wunschkind mit blauen Augen und
schwarzen Haaren? Wo sind die Grenze und wer soll sie festlegen?
Ich bin der Meinung, dass Gentechnik sich positiv auf der Gesellschaft auswirken kann.
Es könnte bestimmt viele Leute helfen und Erkrankungen schneller diagnostifizieren und
heilen. Wir können aber heute nicht absehen, welche Möglichkeiten die Gentechnik in
der Zukunft bieten wird. Deshalb haben viele Leute Angst vor der Entwicklung. Vor allem
müssen wir sicher gehen, dass alle Verfahren gründlich getestet werden, und dass wir
die Wirkung der Gentechnik kennen. Vielleicht wird sie dann akzeptabler.
289
EXAMPLE OF AN ESSAY
[b1004_04]
Leben wir um zu arbeiten oder arbeiten wir um zu leben?
Diese Frage ist schwierig zu beantworten, weil es darauf viele Sache ankommen, zB das
Alter der Person, was für einen Beruf die Person ausübt, ihr Geschlecht usw.
Es wird allgemein angenommen, dass wir arbeiten, um zu leben. Die Mehrheit der Leute
würde wahrscheinlich sagen, dass andere Sachen wichtiger als Arbeit für sie sind: Sie
geben Familiezeit und Freizeit mehr Wert al ihrer Karriere. Darum arbeiten diese Leute
nur, um ihre Betätigungen zu finanzieren, die sie in ihrer Freizeit machen. Sie würden
auch sagen, dass wir auf Arbeit verzichten können, während man ohne andere Sachen
im Leben nicht zurechtkommen würde.
Im Gegenteil, einige Leute würden ohne Arbeit nicht zurechtkommen! Einige Leute
arbeiten, um die fehlende Dinge in ihrem Leben zu ersetzen und andere Leute sagen,
dass sie ohne Arbeit keine Motivation hätten am Morgen aufzustehen.
Heutzutage verändern sich diese Meinungen. Viele Leute studieren jetzt für viele Jahren,
um eine Karriere, die sie genießen zu können zu bekommen und, weil sie so fleißig
arbeiten mussten, um Erfolg zu haben, müssen sie weiter fleißig sein, um erfolgreich zu
bleiben. Wenn man seinen Beruf genießt, kann man leben, um zu arbeiten, zB Medizin
und Rechtsanwälte. Frauen leben zunehmend, um zu arbeiten, weil sie vielleicht fühlen,
dass sie etwas zu beweisen haben. Ich bin der Meinung, dass viele Frauen zwischen 20
und 30 Jahren sagen würden, dass ihre Karriere ihre höchste Angelegenheit ist, obwohl
ich auch denke, dass sich dies für die Mehrheit ändern würde, wenn sie Anfang 30
wären, weil dann viele Frauen eine Familie wollen.
Zum Schluss würde ich sagen, dass es davon abhängt, ob man seinen Beruf genießt.
Wenn man sich freut zur Arbeit zu gehen, ist es möglich zu leben, um zu arbeiten.
Persönlich studiere ich an der Universität, um diese Möglichkeit zu bekommen. Wenn ich
endlich einen Beruf bekomme, dann könnte ich entscheiden, ob ich leben will, um zu
arbeiten, oder ob ich arbeiten will, um zu leben.
290
APPENDIX B
PERIPHRASTIC MODAL EXPRESSIONS
CoAn code
Category description Tag Items
assertive modal expressions
110 purely assertive gewiss, gewisslich, tatsächlich, wirklich 121 evaluative assertive –
positive begrüßenswerterweise,
dankenswerterweise, zum Glück, glücklicherweise, Gott sei Dank, gottlob, günstigerweise
122 evaluative assertive – reservedly positive
immerhin, wenigstens, zumindest
123 evaluative assertive – negative
bedauerlicherweise, leider, unglücklicherweise
124 evaluative assertive – other dimensions
*weise
130 evidence-focused assertive
auf alle Fälle, begreiflicherweise, bekanntermaßen, bekanntlich, bewiesen, bezeichnenderweise, charakteristischerweise, definitiv, eh, erklärlicherweise, erwiesenermaßen, feststehen, fraglos, hundertprozentig, in jedem Fall, logischerweise, leugnen, nachgewiesenermaßen, natürlich, nachweislich, offenkundig, offensichtlich, ohnedies, ohnehin, selbstredend, selbstverständlich, sowieso, Tatsache, unanfechtbar, unleugbar, unzweifelhaft, kein Zweifel, zweifellos, zweifelsohne,
modally qualifying expressions (epistemic)
210 compelling conclusion IS bestimmt, mit Sicherheit, sicher, sicherlich, überzeugt, Überzeugung
220 subjective assumption IV Ansicht, Auffassung, Erfahrung, Ermessen, ich denke, ich glaube, meiner Meinung nach
230 tentative conclusion ISA annehmen, anscheinend, allem Anschein nach, höchstwahrscheinlich, mutmaßlich, nach menschlichem Ermessen, normalerweise, offenbar, schätzungsweise, scheinen, ziemlich sicher, vermuten, vermutlich, aller Voraussicht nach, voraussichtlich, wahrscheinlich, aller Wahrscheinlichkeit nach, größter Wahrscheinlichkeit an Sicherheit grenzender Wahrscheinlichkeit, wohl
240 (tentative) assumption of possibility
IA/IAV Aussicht, *bar, Chance, denkbar, eventuell, möglich, möglicherweise, Möglichkeit, Potential, so Gott will, unter Umständen, vielleicht, vorstellbar
291
250 evidentials IBA/IBS angeblich, gerüchteweise, wie verlautet
negative assertive modal expressions
310 reservedly negative beinahe, eher nicht, fast, kaum, scheinbar, schwerlich, in Zweifel, zweifelhaft
320 emphatically negative ausgeschlossen, aussichtslos, auf keinen Fall, gar nicht, keineswegs, mitnichten, nie, niemals, nirgendwo, nirgends, überhaupt nicht, unausführbar, undenkbar, unmöglich, Unmöglichkeit, unvorstellbar
modally qualifying expressions (deontic)
410/411 duty, compulsion RN/NegRN erforderlich, genötigt, nichts anderes übrig, nicht umhin kommen, notwendig, notwendigerweise, Notwendigkeit, nötig, unabwendbar, unausweichlich, unumgänglich, unumstößlich, unvermeidlich, keine Wahl, verpflichtet, zwingen
420/421 instruction/obligation RF/NegRF angeordnet, befohlen, geboten, gefordert, verfügt, veranlsasst
430/431 suggestion/advice RFV/NegRFV wäre angebracht/besser/ gut/ratsam/richtig
440/441 intention RW/NegRW anstreben, abgeneigt sein, aus sein auf, beabsichtigen, Entschluss, Intention, es leid sein, (keine) Lust, planen, Plan, es satt haben, Trieb, vorhaben, Wille, Wunsch, sich sträuben/zieren /widersetzen, verlangen
450/451 volition RWA/NegRWA hoffen, hoffentlich 460/461 attenuated volition RZ/NegRZ gern, geschmacklos, Hang, hassen,
lieben, (nicht) leiden, Neigung, Schwäche, unmöglich, unausstehlich, unangenehm
470/471 possibility due to circumstances
RMU/NegRMU möglich, Möglichkeit
480/48 possibility due to ability RMF/NegRMF Begabung, beherrschen, Einfluss, Eigenschaft, Erfahrung, Gabe, fähig, Fähigkeit, imstande, Kenntnis, in der Lage,vermögen, zustande bringen
490/491 permission RE/NegRE Befugnis, (nicht) erlaubt, Erlaubnis, (nicht) gestattet, verboten
292
APPENDIX C
DATA TABLES CORRESPONDING TO OVERVIEW CHARTS
Figure 6: Individual modal verb frequencies (per 10,000 words)
Year A Year B Year C KEDS
KÖNNEN 110.8 64.5 95.8 80.2
MÜSSEN 35.5 35.2 46.0 33.8
SOLLEN 27.3 16.5 53.8 23.1
WERDEN 38.4 34.2 32.5 10.0
WOLLEN 28.6 23.8 26.5 10.2
DÜRFEN 2.6 2.4 8.6 6.3
MÖGEN 11.1 12.4 4.3 5.5
Figure 7: Epistemic modal verb frequencies (per 10,000 words)
Year A Year B Year C KEDS
WERDEN 38.4 34.2 32.5 10.0
KÖNNEN/KÖNNTE 25.8 7.0 26.0 13.0
MÜSSEN 0.9 0.8 1.8 1.2
MÖGEN 0.2 0.3 0.2 1.3
SOLLTE 0.0 0.2 0.9 1.7
MÜSSTE 0.0 0.0 0.1 1.0
DÜRFTE 0.0 0.2 0.1 0.7
SOLLEN (evident.) 0.0 0.0 0.3 0.7
WOLLEN (evident.) 0.0 0.0 0.0 0.1
293
Figure 8: Non-epistemic modal verb frequencies (per 10,000 words)
Year A Year B Year C KEDS
MÜSSEN/MÜSSTE 34.5 34.3 43.1 31.6
KÖNNEN/KÖNNTE (circum.) 45.3 32.6 25.5 40.8
KÖNNEN/KÖNNTE (ability) 20.2 10.6 18.7 15.0
SOLLTE 12.6 5.8 25.0 15.2
SOLLEN 14.7 9.9 26.5 5.4
WOLLEN 28.4 23.7 26.5 10.1
MÖCHTE 7.4 9.1 3.4 3.9
MÖGEN 3.5 3.0 0.7 0.2
DÜRFEN 2.6 2.3 8.4 5.6
KÖNNEN (permiss.) 1.1 3.8 2.7 1.2
Figure 14: Frequencies of assertive PMEs (per 10,000 words)
Year A Year B Year C KEDS
purely assertive 2.8 2.6 12.0 5.0
EA-positive 0.7 0.8 0.6 0.1
EA-reservedly positive 0.7 0.6 0.6 3.8
EA-negative 2.8 2.4 2.6 2.6
EA-other dimensions 0.7 0.5 0.6 0.4
evidence-focused assertive 12.4 15.2 17.6 16.0
294
Figure 15: Frequencies of modally qualifying periphrastic expressions (per 10,000 words)
Year A Year B Year C KEDS
compelling conclusion 2.8 1.8 5.0 8.8
tentative conclusion 6.9 14.7 12.4 11.5
subjective assumption 34.2 24.6 10.4 8.0
possibility 10.4 16.2 18.8 10.1
evidential 0.0 0.3 1.5 0.9
Figure 18: Frequencies of non-epistemic PMEs (per 10,000 words)
Year A Year B Year C KEDS
duty/compulsion 1.7 3.3 4.1 3.9
possibility 5.6 7.6 13.0 17.8
ability 0.4 0.6 1.8 3.3
suggestion/advice 0.2 0.0 0.1 0.2
instruction 0.2 0.3 0.2 0.7
volition 0.7 0.9 0.5 0.5
attenuated volition 3.0 5.9 6.8 2.1
affection/liking 1.5 1.4 0.7 1.1
permission 1.5 1.2 6.2 2.4
295
Figure 19: Frequencies of Konjunktiv II context types (per 10,000 words)
Year A Year B Year C KEDS
non-factual argumentation 29.0 20.8 70.1 52.5
conditionals 10.6 9.6 30.0 7.7
indirect speech 7.4 5.3 6.5 6.6
politeness 0.0 1.4 0.6 0.7
others 0.2 0.9 0.8 0.2
errors (indicative required) 4.6 7.7 4.8 0.0
Figure 16: Distribution of Konjunktiv II across context types as percentages of overall Konjunktiv II frequencies
Year A Year B Year C KEDS
non-factual argumentation 56.1 45.5 62.1 77.5
conditionals 20.5 20.9 26.6 11.3
indirect speech 14.2 11.6 5.8 9.7
politeness 0.0 3.0 0.5 1.1
others 0.4 2.0 0.7 0.4
errors (indicative required) 8.8 16.9 4.3 0.0
296
Figure 21: Conditional clause patterns as % of overall frequencies of conditionals
Year A Year B Year C KEDS
P(I) – A(I) 77.8 74.3 62.1 84.0
P(KII) – A(KII) 13.8 15.0 19.2 5.3
P(I) – A(KII) 7.4 9.3 16.3 7.1
P(KII) – A(I) 1.0 0.4 2.3 2.8
other 0.9 0.7 0.0 0.0
297
APPENDIX D
EXAMPLES OF HANDOUTS WITH “REDEMITTEL” (USEFUL PHRASES) AND EXERCISES IN
PREPARATION FOR WRITING ARGUMENTATIVE TEXTS.
Part I: Phrases and exercises used in preparation for written texts