Phraseology and Style in Subgenres of The Novel

Phraseology and
Style in Subgenres
of the Novel
A Synthesis of Corpus
and Literary Perspectives
Edited by
Iva Novakova · Dirk Siepmann
Phraseology and Style in Subgenres of the Novel
Iva Novakova · Dirk Siepmann
Editors
Phraseology
and Style
in Subgenres
of the Novel
A Synthesis of Corpus and Literary
Perspectives
Editors
Iva Novakova Dirk Siepmann
Grenoble Alpes University University of Osnabrück
Grenoble, France Osnabrück, Germany
ISBN 978-3-030-23743-1 ISBN 978-3-030-23744-8 (eBook)

https://doi.org/10.1007/978-3-030-23744-8
© The Editor(s) (if applicable) and The Author(s) 2020

This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by
similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with
regard to jurisdictional claims in published maps and institutional affiliations.
Cover credit: Marina Lohrbach_shutterstock.com
This Palgrave Macmillan imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
The genesis of this book was in a four-year collaborative research project

PhraseoRom1 on the phraseology of contemporary novels, co-funded
by the French National Research Agency (ANR2) and the German
Research Foundation (DFG3). It is one of the few international projects
to truly bring together researchers from both literary studies and lin-
guistics. The book, whose ten chapters report on selected results of this
project, revolves around a detailed analysis and classification of recur-
rent fiction-specific patterns found in fictional genres and their gen-
eral functions, as revealed by sophisticated corpus-driven enquiry. It
focuses both on patterns found in the novel generally and genre-spe-
cific patterns shared by various literary genres. In addition, the book
1https://phraseorom.univ-grenoble-alpes.fr. The PhraseoRom project (2016–2020), led by

I. Novakova (University Grenoble Alpes, France) and Dirk Siepmann (University of Osnabrück,
Germany), brought together 25 researchers from different French and German universities
[University Grenoble Alpes, Sorbonne-Nouvelle Paris 3 University, University of Montpellier
(France), University of Bonn, University of Osnabrück and University of Erlangen-Nuremberg
(Germany)].
2Agence Nationale de la Recherche (ANR).
3Deutsche Forschungsgemeinschaft (DFG).
v
vi Preface
compares and contrasts the stylistic practices encountered in British,

American and French contemporary novels published since the 1950s
and discusses implications these might have for phraseology or literary
translation.
The book lies at the intersection of corpus, computational linguistics,
and stylistics and is resolutely situated within the digital humanities.
It is our hope that it will lend impetus to genre studies by being the
first large-scale project to employ Natural Language Processing (NLP)
and digital stylistics tools to describe literary genres not just in terms
of traditional rhetoric or grammar, but more so as lexico-grammatical
artifacts based on recurrent patterns. Thus, the book is primarily con-
cerned with phraseological aspects of style. Since our aim is to explore
the recurrent features of fictional genres and their general functions, we
rarely consider specific authors or novels individually here, although our
methodology could also serve to identify author-specific lexico-syntactic
patterns.
Generally speaking, our research has produced persuasive evidence
to suggest that the identification of recurrent patterns, often called
“motifs” in the French literature on the subject (Longrée and Mellet
2013; Legallois 2012), enables researchers to better distinguish the spe-
cificities of different literary genres. Viewed from this angle, literariness
arises from the significant over-representation of particular motifs in
specific literary genres, most of which are so discreetly conventional that
the reader may not consciously notice them. Put succinctly, motifs are
genre markers at sentence level. Where genre theory is concerned, the
innovative potential of our approach is in particular due to the avenues
it opens to considering long-standing debates on feature analysis, genre
distinction and the hybridization of genres from a new vantage point
and by deploying a new, interdisciplinary methodology.
Presentation of the Chapters

The chapters of this book are arranged systematically to build up the
picture, starting with general issues and general fiction and then moving
on to comparisons between specific genres.
Preface vii
Chapter 1, written by I. Novakova and D. Siepmann, outlines lin-

guistic approaches to literature, to phraseology and idiomaticity, as well
as new approaches in stylistics and in theories of literary genre for char-
acterizing the recurrent lexico-grammatical patterns in contemporary
fiction. It also summarizes our methodology, presents our corpora and
highlights the book’s innovative features.
Chapter 2 (D. Legallois and S. Koch) proposes an overview of the
term “motif ” which refers to recurring patterns in different disciplines:
folkloristics, narratology, bioinformatics, NLP and linguistics.
Chapter 3 (I. Novakova, D. Siepmann and M. Gymnich) analyzes
the key adverbs and adverbial motifs in English and their functional
equivalents in French novels using the keyword approach (Scott and
Tribble 2006).
Chapter 4 (S. Diwersy, L. Gonon, V. Goossens, M. Gymnich and
A. Tutin) deals with verbs introducing direct speech in English and
French contemporary fiction (especially in the crime, fantasy and
romancesubgenres).
Chapter 5 (F. Grossmann, M. Gymnich and D. Siepmann) investigates
the expressions related to alcohol and tobacco consumption in post-war
English and French fiction from a corpus-stylistic and corpus-driven van-
tage point.
Chapter 6 (L. Gonon and O. Kraif ) compares the “fiction words”
(neologisms) in French and American science fiction during the 1990s
to determine to which extent two different literary traditions may share
a common background of fictional references, mixing elements that
come from various “xenoencyclopedias” (Saint-Gelais 1999).
Chapter 7 (V. Goossens, C. Jacquot and S. Dyka) distinguishes
between the two fictional genres of science fiction and fantasy through
an original semantic and stylistic classification of the recurrent narrative
patterns and related motifs in the corpora.
Chapter 8 (J. Sorba, L. Gonon, S. Dyka and V. Goossens) considers
the discursive functions of motifs generated by the expressions écrire/lire
une lettre, un roman, write/read a letter, a novel in general contemporary
fiction in comparison with a corpus of crime novels.
viii Preface
Chapter 9 (S. Dyka, L. Fesenmeier and M. Gymnich) studies the

motifs generated by the structures “dans un état de NP/in a state of NP”
from functional and stylistic points of view in six subgenres: general,
crime, romance, fantasy, historical and science fiction.
Chapter 10 (I. Novakova and D. Siepmann) outlines the most salient
results of our research while emphasizing the interdisciplinary approach
applied in differentiating contemporary fiction subgenres. It also points
out the avenues the book opens to fruitful future research in the digital
humanities.
Grenoble, France Iva Novakova

Osnabrück, Germany Dirk Siepmann
References
Legallois, Dominique. 2012. “La Colligation: autre nom de la collocation
grammaticale ou autre logique de la relation mutuelle entre syntaxe et
sémantique?” Corpus 11. http://corpus.revues.org/2202.
Longrée, Dominique, and Sylvie Mellet. 2013. “Le Motif: une unité
phraséologique englobante? Étendre le champ de la phraséologie de la
langue au discours.” Langages 189: 68–80.
Saint-Gelais, Richard. 1999. L’Empire du pseudo: Modernités de la science-fic-
tion. Québec: Nota Bene.
Scott, Mike, and Christopher Tribble. 2006. Textual patterns: Key words and
corpus analysis in language education. Amsterdam: Benjamins.
Acknowledgements
We owe a great debt of gratitude to Professor Marion Gymnich

(University of Bonn, Germany) for her invaluable aid in re-reading
the various chapters of this book. We sincerely appreciate, too, Denise
Burkhard’s (University of Bonn, Germany) meticulous work in pre-
paring the manuscript for publication, and express our thanks to her.
Finally, we are indebted to Henry Randolph (Tek Ryder Translations,
California, USA) for his skilful editing and proofreading of the draft
chapters.
ix
Contents
1 Literary Style, Corpus Stylistic, and Lexico-Grammatical

Narrative Patterns: Toward the Concept of Literary Motifs 1
Iva Novakova and Dirk Siepmann
2 The Notion of Motif Where Disciplines Intersect:

Folkloristics, Narrativity, Bioinformatics, Automatic
Text Processing and Linguistics 17
Dominique Legallois and Stefan Koch
3 Key Adverbs and Adverbial Motifs in English Fiction

and their French Functional Equivalents 47
Iva Novakova, Dirk Siepmann and Marion Gymnich
4 Speech Verbs in French and English Novels 83

Sascha Diwersy, Laetitia Gonon, Vannina Goossens,
Marion Gymnich and Agnès Tutin
xi
xii Contents
5 Alcohol and Tobacco Consumption in English

and French Novels Since the 1950s:
A Corpus-Stylistic Analysis 115
Francis Grossmann, Marion Gymnich
and Dirk Siepmann
6 French and American Science Fiction During

the Nineties: A Contrastive Study of Fiction Words
and Phraseology 151
Laetitia Gonon and Olivier Kraif
7 Science Fiction versus Fantasy: A Semantic

Categorization and its Contribution to Distinguishing
Two Literary Genres 189
Vannina Goossens, Clémence Jacquot and Susanne Dyka
8 Reading and Writing as Motifs in English

and French General Fiction 223
Julie Sorba, Laetitia Gonon, Susanne Dyka
and Vannina Goossens
9 Dans un état de NP and in a state of NP: Bridging

the Syntagmatic Gap in English and French Fiction 251
Susanne Dyka, Ludwig Fesenmeier and Marion Gymnich
10 Towards an Interdisciplinary Approach for

Differentiating Contemporary Fiction Subgenres 279
Appendix A: Corpora 287
Appendix B: Discursive Functions 291
Index 295
Notes on Contributors
Sascha Diwersy is associate professor of French linguistics in the

Department of Linguistics and the Praxiling research laboratory at Paul
Valéry Montpellier 3 University, France. His main research interests are
in corpus linguistics, syntax, semantics, lexicology, text linguistics and
the study of language varieties. He has authored a textbook on colloca-
tion, corpus-driven lexico-grammatical analysis and contrastive studies
(Kookkurrenz, Kontrast, Profil) and has written some 30 articles. His is
the (co-)author of several large corpora in different languages (French,
German, Spanish, English, Russian) and the (co-)developer of two cor-
pus analysis platforms (EmoBase/Lexiscope, BTLC.PrimeStat/Varitext).
Susanne Dyka currently works at the Institute of English and
American Studies at Osnabrück University and the Institute of
Anglistics at Friedrich-Alexander-University Erlangen-Nuremberg. Her
main research interests are in lexicography and phonology. As a member
of the PhraseoRom project team she specializes in the semantic classifi-
cation of RLTs.
Ludwig Fesenmeier is professor of French and Italian Linguistics
in the Institute of Romance Studies at Friedrich Alexander University
xiii
xiv Notes on Contributors
Erlangen-Nuremberg, Germany. His main research interests are in

(historical) Romance text linguistics, historical syntax of the Romance
languages as well as Romance variational linguistics. He has co-edited
several volumes and written two other books and more than 70 articles
and reviews.
Laetitia Gonon is an associate professor of French language and sty-
listics at the University Grenoble Alpes, France (UGA, UMR5316 Litt
& Arts). She specializes in the style of crime news items and popular
fiction from the nineteenth century to the present day, particularly their
phraseology. She is currently interested in Anglicisms in French novels
(Flaubert, but also contemporary French bestsellers, detective novels
and sentimental novels).
Vannina Goossens is a postdoctoral fellow at the University Grenoble
Alpes and a member of LIDILEM, as part of the Franco-German ANR-
DFG PhraseoRom project. Her work is in the fields of semantics and
corpus linguistics with focus on studying the semantic structuring of
lexicon and lexical units, polysemy, polylexical sequences and their role
in structuring discourse and literary genres using statistical and texto-
metric methods.
Francis Grossmann is professor emeritus of Linguistics at the
University Grenoble Alpes, France, and a member of the LIDILEM
Research Team. His research has focused in recent years on analysing
the phraseological dimensions of scientific discourse, the lexicon of
emotions and the discursive markers of reported speech. His research
interests further include lexical learning at different levels of education.
He has (co)-authored some fifteen books or entire journal issues. He has
also co-directed several research projects and his work has been pub-
lished in over 70 peer reviewed publications as well as book chapters
and conference proceedings.
Marion Gymnich is professor of English Literature and Culture at the
University of Bonn, Germany. Her research interests include British
literature and culture from the nineteenth century to the present, nar-
rative theory, and genre theory. She has (co-)authored six books, (co-)
edited 14 books and has published more than 100 articles.
Notes on Contributors xv
Clémence Jacquot is an associate professor of linguistics and stylis-

tics at the University Paul-Valéry Montpellier 3 and a researcher at the
Praxiling laboratory (UMR 5267). She is a specialist in stylistics and,
during her Ph.D. and as a postdoc in Digital Humanities, did work on
the tools used to interpret literary texts. She is currently interested in
the differences between the narrative subgenres and their description
using digital stylistics tools.
Stefan Koch is a postdoctoral fellow at the University of Erlangen-
Nuremberg (Germany) in the binational PhraseoRom project. He has
to his credit publications on Spanish scriptology, Spanish and Italian
historical linguistics as well as Italian morphology, Romance and general
typology.
Olivier Kraif is professor at the University Grenoble Alpes teaching
in the field of Computer Science and Natural Language Processing.
He has been a researcher in LIDILEM (Laboratoire de LInguistique
et DIdactique des Langues Étrangères et Maternelles) since 2002. He
works in the field of text corpora processing with an emphasis on mul-
tilingual corpora (comparable as well as parallel). His research aims
include developing tools and techniques for investigating linguistic phe-
nomena from lexicon, phraseology, contrastive analysis and translational
studies perspectives.
Dominique Legallois is professor of French linguistics at Sorbonne-
Nouvelle Paris 3 University, France. His main research interests are in
corpus linguistics, discourse analysis and construction grammar. He has
(co-)authored two textbooks on construction grammar (Constructions
in French) and text linguistics (The Grammar of Genres and Styles: From
Discrete to Non-discrete Units) and has written more than 60 articles and
reviews.
Iva Novakova is professor of French and Contrastive Linguistics at
the University Grenoble Alpes and a member of LIDILEM research
team. Her research has focused in recent years on phraseology and
corpus linguistics. She has published more than 80 works in the field
of the emotional lexicon, on the phraseology in literary language, on
causative constructions and on time-modes-aspects. She co-led the
xvi Notes on Contributors
French-German Emolex (2010–2013) research project EMOLEX on

the emotional lexicon in five European languages. Currently, she is
principal investigator, with Dirk Siepmann, of the PhraseoRom project
(2016–2020), whose main objective is studying the extended phrase-
ology and specific textual motifs of contemporary French, English and
German novels.
Dirk Siepmann is professor of English language teaching at the
Institute of English and American Studies of Osnabrück University,
Germany. His main research interests are in lexicology and lexicography,
translation studies and language teaching. He has (co-)authored two
textbooks on academic writing (Writing in English and Wissenschaftliche
Texte auf Englisch schreiben), and written ten other books and more than
80 articles and reviews. Currently, he is principal investigator, with Iva
Novakova, of the PhraseoRom project (2016–2020).
Julie Sorba is an associate professor in Linguistics at the University
Grenoble Alpes (LIDILEM and UMR5316 Litt & Arts Research
Centers), France. She defended her Ph.D. thesis in comparative linguis-
tics of ancient languages. In recent years, she has pursued her academic
research interests along two main tracks: lexicology and especially phra-
seological phenomena in broad corpora of French contemporary—lit-
erary/non-literary—texts using the tools of corpus linguistics. She has
written more than 40 articles and reviews. She is the co-editor of LIDIL
a Journal of Linguistics and Language Education.
Agnès Tutin is professor in French Linguistics at the University
Grenoble Alpes, France, participates in the LIDILEM research team,
and heads up the Masters in Linguistics program. Her main research
interests are in lexical semantics, phraseology and corpus linguistics. In
her recent work, she has focused on phraseological routines in academic
discourse and spoken corpora.
Abbreviations
CONT Contrast corpus

CRIM Crime fiction
DF Discursive functions
DISP Dispersion
EN English
FR French
FW Fiction words
FY Fantasy novels
GEN General fiction
HIST Historical novels
LIT Literary corpus
LLR Log-likelihood ratio
LSCs Lexico-syntactic constructions
NLP Natural language processing
NP Nominal phrase
RLT(s) Recurrent lexico-syntactic tree(s)
ROM Romance novels
SF Science fiction novels
xvii
List of Figures
Chapter 2
Fig. 1 Genome sequences 28
Fig. 2 Verbal valence as elements of the motif
(based on Čech et al. 2017, 16) 30
Fig. 3 Frequency of the motifs (based on Čech et al. 2017, 16) 30
Fig. 4 An example of the RLT pousser la porte (push the door open) 36
Chapter 4
Fig. 1 CA applied to the data set obtained by cross-tabulating verb
class and the combination of language with genre (The verb
class is shown in red, the combined category of genre
and language in blue. Circle size represents contribution) 94
Fig. 2 Vocabulary growth curves for the verb slot of the speech tag
construction in the English and French samples 97
Chapter 5
Fig. 1 Most frequent types of alcohol in the English and French
sub-corpora (n English = 8046; n French = 6837) 141
xix
xx List of Figures
Fig. 2 “Other” types of alcohol in the English and French

sub-corpora, n = 229 in the English corpus, n = 239
in the French corpus 141
Fig. 3 Take a sip in the English corpus before and after sorting,
n = 303 before sorting; 202 after sorting 143
Fig. 4 Boire une gorgée in the French corpus before and after
sorting, n = 207 before sorting; 113 after sorting 144
Fig. 5 Sip of wine and gorgée de vin in the English
and French corpora 145
Chapter 6
Fig. 1 Distribution of the selected words in the French
corpus (n.b. the OTH label refers to words that have
not been categorized) 162
Fig. 2 Distribution of selected words in the English corpus 163
Fig. 3 Distribution of types of formation for the selected words
in the French corpus 165
Fig. 4 Distribution of types of formation for the selected words
in the English corpus 166
Chapter 9
Fig. 1 Recurrent Lexico-Syntactic Tree for dans un état de
NP in the French sub-corpora 255
Fig. 2 Recurrent Lexico-Syntactic Tree for in a state of
NP in the English sub-corpora 255
List of Tables
Chapter 2
Table 1 The Dragon-Slayer as a motif chain and its equally
valid story variants (based on Ofek et al. 2013, 3) 28
Table 2 Textual motif “discovery of a body” (based on Muryn
et al. 2016, 9–11) 37
Chapter 3
Table 1 Quantitative differences between different types
of adverbs in English and French 49
Table 2 English and French key manner adverbs
with potential equivalents 53
Table 3 Quantitative differences between selected candidates
for equivalence (occurrences per one million words) 54
Table 4 Distribution of manner adverbs by verb semantics
(log dice >5) 56
Table 5 Natural French equivalents of English adverbs 59
Table 6 Frequency of key adverbs in translation and comparable
corpora revisited 78
xxi
xxii List of Tables
Chapter 4
Table 1 Corpora used for the study 88
Table 2 Occurrences of direct speech 89
Table 3 Frequencies of verb classes cross-tabulated
by genre combined with language 93
Table 4 Contrastive specificities of verb classes by language 95
Table 5 Contrastive specificities of verb classes by genre
in the French data set 96
Table 6 Contrastive specificities of verb classes by genre
in the English data set 96
Table 7 Frequencies of verb classes cross-tabulated by language 110
Table 8 Frequencies of verb classes cross-tabulated by genre
for English 111
Table 9 Frequencies of verb classes cross-tabulated by genre
combined for French 111
Chapter 5
Table 1 Most frequent RLTs related to the cigarette script
in the English sub-corpus 126
Table 2 Most frequent RLTs related to the cigarette
in the French sub-corpus 132
Table 3 Most frequent RLTs related to the consumption
of alcohol in the English sub-corpus 137
Table 4 Most frequent RLTs related to the consumption
of alcohol in the French sub-corpus 140
Chapter 6
Table 1 Quantitative description of the corpora 153
Table 2 Semantic classes derived from the selected fiction words 161
Table 3 Comparative distribution according to POS 163
Chapter 7
Table 1 French and English science fiction and fantasy corpora 191
Table 2 Cumulative thresholds and number of RLTs
for each language and genre 193
List of Tables xxiii
Table 3 Semantic dimensions and values used

in the PhraseoRom project 195
Table 4 Classification of patterns by semantic dimension
for French science fiction and fantasy 201
Table 5 Classification of RLTs by semantic dimension
for English science fiction and fantasy 202
Chapter 8
Table 1 Specificities of the recurrent lexico-syntactic patterns
of our study in the GEN corpora 228
Table 2 The lexical and grammatical collocates of the eight patterns 230
Table 3 Syntagmatic variations of the article across the eight LSCs 232
Table 4 The syntagmatic variations of the noun across the eight LSCs 233
Table 5 The syntagmatic variations of the verb across the eight LSCs 233
Table 6 The paradigmatic variations across the noun in the eight LSCs 235
Chapter 9
Table 1 Frequency of dans un état de NP and in a state of
NP with statistically relevant collocates and number
of statistically relevant collocates (LLR ≥10.83) 256
Table 2 Frequency of en état de NP 257
Table 3 Statistically significant collocates of in a state of
NP sorted by word classes 258
Table 4 Statistically significant collocates of dans un état de
NP sorted by word classes 258
Table 5 Statistically significant verb collocates of dans un état de
NP (LLR ≥10.83) 259
Table 6 Statistically significant verb collocates of in a state of
NP (LLR ≥10.83) 260
Table 7 Statistically significant noun collocates of dans un état de
NP (LLR ≥10.83) 262
Table 8 Statistically significant noun collocates of in a state of
NP (LLR ≥10.83) 263
Table 9 Statistically significant adjective collocates
of dans un état de NP (LLR ≥10.83) 264
Table 10 Statistically significant adjective collocates
of in a state of NP (LLR ≥10.83) 265
xxiv List of Tables
Appendix A
Table 1 Size of the entire comparable corpora 288
Table 2 Size of the samples in the literary corpora (LIT)
(Samples of literary corpora (LIT) versus reference
corpora [CONT] [cf. Table 3]) 289
Table 3 Size of the contrast (non-literary) corpora (CONT) 290
Table 4 Size of the parallel corpora 290
1
Literary Style, Corpus Stylistic,
and Lexico-Grammatical Narrative
Patterns: Toward the Concept
of Literary Motifs
1 Introduction
In this chapter, Section 2 opens with an outline of the linguistic
approaches to literature, to phraseology and idiomaticity, as well as new
approaches in stylistics and in theories of literary genre, to character-
ize the recurrent lexico-grammatical patterns in contemporary fiction.
In Sect. 3, we summarize our methodology and present our corpora.
Section 4 highlights the book’s innovative features. This section also
defines what sets the patterns called “motifs” apart from other types
of phraseological units and how the present work advances research in
linguistics and literary studies.
I. Novakova (*)
Grenoble Alpes University, Grenoble, France
e-mail: iva.novakova@univ-grenoble-alpes.fr
D. Siepmann
University of Osnabrück, Osnabrück, Germany
e-mail: dirk.siepmann@uni-osnabrueck.de
© The Author(s) 2020 1
I. Novakova and D. Siepmann (eds.),
Phraseology and Style in Subgenres of the Novel,
https://doi.org/10.1007/978-3-030-23744-8_1
2
I. Novakova and D. Siepmann
2 Scientific Background
2.1 Linguistic Approaches to Literature
Previous research in stylistics (e.g. Barthes 1966; Leech and Short 2007),
corpus stylistics (Stubbs 2005; Fischer-Starcke 2010; Mahlberg 2013) and
textometry (Brunet 1981)1 concentrated on characterizing the style(s) of
various authors (e.g. Flaubert, Proust, Dickens, Austen). It showed that
the bulk of the theoretical literature focuses on recurrent schemas (e.g.
Todorov 1980; Lits 2011) found in their novels. On the other hand,
research is scarce when it comes to fiction-specific lexico-grammatical pat-
terns based on large corpora, which the present volume centers on. Our
study first differentiates these patterns before proceeding to distinguishing
them from other types of phraseological units.
While some literary scholars (e.g. Attridge 2004) and the general
public tend to confer a special status on the language of literature, lin-
guists generally agree that “literary language is not special or different,
in that any formal feature termed ‘literary’ can be found in other dis-
courses” (Burton and Carter 2006, 273). Countering the formalist claim
that “defamiliarization” or “foregrounding” (Mukařovský 2014, 43)
is the essence of literature and literary language, a strong case has been
made that many works of literature contain “ordinary language” or have
their “roots in everyday uses of language” (Leech 2014, 5–6). This has
led to attempts at capturing the specificity of literary language in func-
tional terms, using criteria such as medium-dependence, displaced inter-
action, and polysemy (Burton and Carter 2006, 272) or the “duplicity”
(Scholes 1982, 23) of the various factors involved in the communication
process (e.g. the difference between author and narrator).
If we adopt this view, the subjective impression of “literariness”
(литepaтypнocть, Jakobson 1921) conveyed by even the shortest pas-
sage of imaginative prose would merely be an incidental phenome-
non subordinate to the unfathomable rules of the artistic craft. Yet,
1[Textometry is an approach that has been developed primarily in France since the 1970s. It
makes use of a large range of linguistically significant and mathematically sound computations
for the analysis of textual data];
1 Literary Style, Corpus Stylistic, and Lexico-Grammatical …
3
significantly, this assumption of a functional difference without a formal

correlative—a kind of linguistic epiphenomenon—is without parallel or
precedent in linguistics. It may simply be due to the fact that literary
language or the language of a particular literary genre have never been
examined in their entirety. As mentioned earlier, the investigations have
usually focused on a particular author’s style or individual texts as well
as on the stylistic devices handed down from Greek antiquity. By turn-
ing both literary and linguistic traditions on their heads as it were, we
are assuming that literariness does not primarily reside in any stylistic
features peculiar to literary texts (e.g. metaphor, irony), but rather in
their adherence to genre-internal conventions of idiomaticity. In other
words, it is the statistically significant over-representation of particular
general-language features that creates the subjective impression of liter-
ariness. It will be argued here that the essence of imaginative writing is
not idiosyncracy or originality but a certain unobtrusive conventional-
ity common to all instantiations of a particular literary genre during a
given period, regardless of their literary status. Any attempt to pin down
the notion of literariness must therefore consider the habitual “norm”
(Coseriu 1975, 85–88) or the “idiom principle” (Sinclair 1991, 113)
underlying specific genres if we are to succeed in determining how the
“creativity principle” (Siepmann 2011, 68) and the “open-choice prin-
ciple” (Sinclair 1991, 175) variously operate in specific texts. Moreover,
Stubbs and Barth (2003, 79) demonstrated that “text types are distin-
guished by lexical and grammatical patterns.” They found that fiction,
for example, is “characterized by a verbal style, by past tense verb forms
and by frequent vocabulary from the lexical fields of saying, looking,
thinking and wanting.” However, the study in question, apart from
being based on a small corpus, was limited by its exclusive focus on
the 200 most-frequently occurring multi-word strings in each genre.
The results showed little evidence of lexical units of meaning capa-
ble of significantly impacting the reader’s conscious perceptions of
texts. Similarly, Biber et al. (1999) identified a number of general fic-
tion-specific grammatical features, such as the absence of participial rel-
ative clauses (606) or the frequent use of double genitives (309). Biber
(1988) and Conrad and Biber (2001) also provided detailed multidi-
mensional analyses of register variation linking situational characteristics
4
to linguistic features and their functions but in which they failed to

consider fiction-specific keywords or multi-word units.
2.2 Approaches to Idiomaticity and Phraseology
As we have just seen, while there is a dearth of studies on the lexi-

co-grammatic nature of literary texts, many researchers have examined
idiomatic combinations in other genres such as journalistic and scientific
texts (Sinclair 1991, 2004; Hunston and Francis 2000; Hoey 2005; de
Beaugrande 2005). While there is currently no consensus among authors
on the resulting profusion of labels used to identify idiomaticity—
such as “extended units of meaning” (Sinclair 2004), “constructions”
(Goldberg 1995), “collostructions” (Stefanowitsch and Gries 2003),
“collocations” (Hausmann 1979; Mel’čuk et al. 1995; Siepmann 2005),
“lexical bundles” (Biber et al. 1999), “sequential patterns” (Quiniou et al.
2012), and “multi-words expressions” (Steyer and Brunner 2014)—there
is nevertheless a growing convergence among these different approaches
and labels that dispenses with the distinction between a grammar com-
posed of rules and a lexicon consisting of words and phrases.
Thus, the Neofirthian approach, whose most accomplished propo-
nent is undoubtedly Hoey (2005), advocates for a grammatical lexicon
containing both grammatical combinations (“colligations”: e.g. GN + to
be + about + V-ing ) and lexical combinations (“collocations” clear motor-
way ). This theory uses the concept of collocational “nests,” where the
meaning is not compositional (say a word => say a word against => won’t
say a word against; Hoey 2005, 11). In addition, Hoey completes the
description of “local” lexico-syntactic relationships, i.e. those internal to
the sentence, by analyzing lexico-syntactic relationships in larger textual
units called “textual colligations.” However, his study is based on a small
number of English lexemes and needs to be expanded to larger corpora
or performed from a cross-linguistic perspective.2
2Lexical priming theory was used for the analysis of the emotion lexicon in five European lan-
guages based on large multilingual corpora (see among others Novakova and Melnikova 2013;
Novakova 2015).
5
Construction grammars, on the other hand, treat language as an

inventory of form/meaning pairs extending across a continuum,
from lexicon to grammatical structures via idiomatic sequences (e.g.
Goldberg 1995; Fillmore et al. 1988; Croft 2001). An obvious rela-
tionship exists between contextualism and certain construction gram-
mars such as the “collostructions” of Stefanowitsch and Gries (2003).
While the former starts from general constructions such as the dit-
ransitive construction to identify its associated lexis, the latter instead
start from individual lexemes (e.g. give) to arrive at general syntactic
patterns.
The strong convergence of these different approaches led to a shift
in research focus from the fixed sequences of traditional phraseology
(routine formulas, proverbs and binary collocations (Hausmann 1979;
Mel’čuk et al. 1995) to all kinds of “pragmatemes” (Feilke 1996) and
“extended” lexical units (Sinclair 2004). Three particularly promising
concepts in this respect are “collocational frameworks” (Renouf and
Sinclair 1991), “motifs” (Legallois 2006, 2012; Longrée and Mellet
2013) and “probabemes” (Herbst and Klotz 2003). The latter are
polylexical sequences highly likely to be used by speakers in particular
situations (e.g. “il n’avait pas de mots assez durs pour Vinf ” vs. “il avait
des mots durs pour Vinf ”; Hausmann 2007, 136).
In this volume, we venture beyond binary collocations by analyz-
ing recurrent polylexical units automatically extracted from our con-
temporary fiction corpora. As such, we firmly position ourselves in the
so-called “extended phraseology” framework (cf. Legallois and Tutin
2013), which includes a wide range of study objects, from collocations
through paremiology or various pragmatics or discursive sequences to
polylexical units. More generally, both the “continental” approach to
phraseology and British contextualism (Sinclair 2004; Hoey 2005;
Biber 2009) converge by no longer considering phraseology as a mar-
ginal phenomenon but rather as an essential structural element of
human language. Several of the contributions in this volume aim spe-
cifically at clarifying how polylexical phraseological units form literary
motifs as well as their functioning both linguistically and discursively in
modern fiction.
6
2.3 Stylistics and Theories of Literary Genre
Textual genres are currently the subject of interdisciplinary studies in

linguistics, literary theory, stylistics, and lexicometric analysis. In the
theory of literature, several works on textual genres authored by, for
example, Gymnich et al. (2007), Zymner (2003), Frow (2005), Duff
(2000), and Monte and Philippe (2014) can be cited here. In addition,
Adam (2005) proposed dealing with the question of text categorization
in terms of dynamics, the effects of genericity and generic tensions: for
him, a text is always in production, as in reception/interpretation, in
relation with one or more genres. This content-based approach, which
still needs to be complemented by linguistics and digital stylistics, opens
new perspectives on the theory of genres (see also Beauvisage 2001;
Rastier 2011).
Furthermore, genre theory has traditionally tended to draw a rela-
tively clear distinction between so-called highbrow and lowbrow fic-
tion, that is, between “popular genres” and “literary fiction” (cf. Frow
2005, 128). This was chiefly done using criteria based exclusively on the
content of the genres in question, such as the description of emotional
states in a romance. Other approaches in genre theory have highlighted
genres as being “rooted in institutional infrastructures” (Frow 2005,
128); that is, generic distinctions are perpetuated by institutions such as
publishing houses and booksellers. Without question, such institutions
are capable of promoting genres through various marketing strategies,
say by using covers that readers perceive as typical of a particular genre.
Still, the reader will ultimately always fall back on genre expectations
when it comes to the content. For content to be deemed characteristic
of a genre, it arguably also has to be couched in a genre-appropriate
language that satisfies readers’ expectations.
Traditionally, both content and language of lowbrow or popular genres,
such as the crime news item, serial novels (Gonon 2012, 2015), detective
stories3 (Eisenzweig 1983), and fantasy fiction, have been thought of as
3L’Affaire Lerouge by E. Gaboriau (1866) is widely accepted as the first French crime novel.
7
shaped by “the formulaic and the conventional” (Frow 2005, 2). In recent
decades, however, in the wake of Postmodernism, the boundaries between
popular and “literary” fiction have progressively blurred, and numerous
factors “have contributed to closing the gap to a certain extent between
highbrow literature and popular culture” (Nünning and Nünning 2018,
30). Genres like crime fiction, historical novels, fantasy, romance and sci-
ence fiction are still intact and thriving. What has changed is the grow-
ing number of well-known “literary” or “highbrow” fiction writers who
have adopted conventions of “generic fiction” in penning their novels.
While Nobel Prize winner Kazuo Ishiguro’s novel When We Were Orphans
(2000) may still be a far cry from formulaic crime fiction, it neverthe-
less clearly draws on various traits proper to detective fiction. In France, a
multi-awarded and acclaimed author like Jean Echenoz also plays with the
rules of historical (14, 2012) or spy fiction (Envoyée Spéciale, 2016). Many
further examples of this kind could be cited.
Given these sorts of dynamics, literary genres tend to exhibit an
extreme literary and stylistic heterogeneity: they range from works that
are definitely categorizable as “lowbrow fiction” (such as the Mills and
Boon/Harlequin romances) to intellectually demanding novels written
by famous authors. Therefore, this categorization is ripe for reassessment
using the tools of modern digital stylistics. The digital-stylistic approach
provides a new type of quantitatively based evidence (see Herrmann
et al. 2015). Consequently, large corpora are changing the stylistic
studies paradigm by offering new heuristic tools to put subgenres into
literary and stylistic categories and revisit the controversial distinction
between highbrow and lowbrow fiction (see Boyer 2008).
3 Corpora and Methodology

In all chapters of this volume, we adopt a corpus-driven methodology4
applied to large English and French literary corpora compiled in the
course of the PhraseoRom project. The corpora consist of contemporary
4For the distinction between corpus based and corpus driven approaches, see Tognini-Bonelli (2001).
8
novels, subdivided into the six subgenres of crime, science fiction,

fantasy, romance, historical and general fiction. Complementing them
is a parallel corpus containing novels that have been translated from
English into French or vice versa (see Appendix A). The corpora were
automatically parsed to carry out lemmatization and part-of-speech tag-
ging as well as marking syntactic dependencies between words. We used
two different tools, Connexor (Tapanainen and Järvinen 1997) for the
French texts, and XIP (Aït-Mokhtar et al. 2002) for the English texts, to
extract lexical units (pivots) that combine with other words and these in
turn provided two different methods of accessing textual motifs:
– the RLTs5 (Recurrent Lexico-syntactic Trees) method, integrated into

Lexicoscope (Kraif 2016; see Chapter 2 by Legallois and Koch in this
volume);
– the keyword approach tool (Scott and Tribble 2006), integrated into
Sketchengine (Kilgarriff et al. 2014; see Chapter 3 by Novakova et al.
in this volume)
The RLTs methodology automatically extracts statistically significant

nonlinear fiction-specific constructions from the literary corpora (LIT)
that are then compared to newspaper, scientific texts and TED talks
that we constituted as our reference corpus (CONT) (see Appendix A).
Based on the complex pivot concept, this iterative extraction method
provides recurrent expressions of length n around a given pivot that
vastly improve on simple frameworks consisting of repeated segments
or lexical chunks. They represent real syntactic sub-trees, which can be
realized in different ways on the surface in the texts. The RLTs are then
automatically clustered according to specificity, frequency, and disper-
sion criteria and analyzed syntactically, semantically, and discursively.
The second approach using keywords is employed in analyzing fic-
tion-specific adverbial constructions in English and French (see Chapter 3
by Novakova et al. in this volume). The keyword in its statistical sense
5The abbreviation RLTs (Recurrent Lexico-syntactic Trees) is the English equivalent for ALR
(Arbres Lexico-syntaxiques récurrents; Kraif 2016).

9
is by now a well-established concept in corpus linguistics: A word must

exceed a minimum frequency level according to the statistical index log
likelihood. Keyword searches capture two kinds of data: (a) what the text
or texts being examined are about (“aboutness indicators”) and (b) what
the predominant word choices are in a stylistic sense (“stylistic indica-
tors”). The corpus-stylistic study of literary language is mostly concerned
with the latter.
We applied a complementary approach to analyzing the speech verbs
in French and English novels (see Chapter 4 by Diwersy et al. in this
volume). Here, the data sets were processed using what essentially are
canonical methods in the tradition of French textometry (cf. Lebart
et al. 1998) whose focus is on contrastive word specificities and corres
pondence analysis. The contrastive specificities method (Lafon 1980)
operates roughly on the same principle as the keyword method in the
British tradition of corpus linguistics but, unlike the latter, runs its cal-
culations using hypergeometric distribution.
4 Innovative Features of the Book

In this section, we highlight what we believe is the unique contribution
our research makes to stylistics by introducing the notion of “motif ”
to bridge the gap between linguistics and literary studies. Our intent
here is not to preempt the detailed discussions found in the remaining
chapters of this book, but to preliminarily sketch what we mean by
this relatively new idea. To begin with, we do not intend for the term
“motif ” to be taken as referring to elements with symbolic significance
that recur in a story, as has traditionally been the case in literary studies.
Instead, in our usage, “motif ” denotes a sequence of either continuous
or discontinuous units that may combine different types of elements
(particular word forms, lemmas and/or morpho-syntactic categories,
collocations). Our motifs thus
– display lexico-syntactic regularities and variations at the syntagmatic

and paradigmatic levels while simultaneously performing particular
discursive/narrative functions. They are therefore recurrent linguistic
10
units that can be described at the levels of lexico-grammar, semantics

and pragmatics/discourse (Longrée and Mellet 2013; Legallois 2012);
– furnish a link between linguistics and literary studies to the extent
that they collaborate in the construction of scripts and schemas; and
are situated—unlike traditional literary motifs—where social scripts
and fictional scripts (see also Baroni 2002, 2007) intersect.
Motifs as we understand them cannot be identified by fully auto-

matic procedures but instead require the linguist and the literary
scholar to make a judgment. The automatic extraction of recurrent
lexico-syntactic trees (RLTs) or of keywords represents the first step in
our general procedure and should be thought of as merely a heuristic
process useful for detecting potential candidate motifs. To take a sim-
ple example, the sequence “Poss Det + N (thoughts/meditations…)
+ were + interrupted/cut short by N/as clause” (e.g. His thoughts
were interrupted by the sound of a key turning in the lock ) may qualify
as a motif to the extent that it combines different types of elements
(grammatical: the noun thought ) and displays paradigmatic variations
(lexical: interrupt/cut short; thoughts/meditations/…) (Siepmann
2015). At the same time it also performs a particular narrative func-
tion: transitioning between an internal and an external narrative per-
spective and beginning a new narrative segment.
The notion of motif lets us fill in the missing link between “ macro-level”
notions of script or schema that have traditionally been used in cognitive
narratology (e.g. study of the plot, isotopies) and the “micro-level” ele-
ments that go into making up the script (specific phraseological recur-
rences). A straightforward example is afforded by what might be called the
“cigarette script,” which, at its most basic, involves a sequence of actions
such as lighting up, inhaling, exhaling, blowing smoke, etc. (for more
details, see Chapter 5 by Grossmann et al. in this volume). A recurrent
motif found in this script type is Pronoun/Noun He + V lighted + DET
a + N cigarette + PP (prepositional phrase), where the PP “slot” is commonly
filled by elements of the type “locative” (e.g. on the way down from the
hotel), “source” (from a packet of Gauloises), “mode of lighting” (from the
burning end of another), or “manner” (with trembling hands).
11
The present volume is intended for linguists, translation and literary

scholars, corpus stylistics scholars and lexicologists, as well as the lay
reader outside academia. We believe that the book may also have a cer-
tain appeal for creative writing teachers and students because it explores
the interrelationships between style and substance or between narrative
intent and formulation, while also providing a wide-ranging catalogue
of specific linguistic patterns that aspiring writers can use or adapt to
their own purposes.
References
Adam, Jean-Michel. 2008 (2005). La Linguistique textuelle: introduction à l’an-
alyse textuelle des discours. Paris: Armand Colin.
Aït-Mokhtar, Salah, Jean-Pierre Chanod, and Claude Roux. 2002. “Robustness
Beyond Shallowness: Incremental Deep Parsing.” Journal of Natural
Language Engineering 8 (2/3): 121–44.
Attridge, Derek. 2004. Peculiar Language: Literature as Difference from the
Renaissance to James Joyce. London and New York: Routledge.
Baroni, Raphaël. 2002. “Le Rôle des scripts dans le récit.” Poétique 129: 105–26.
Baroni, Raphaël. 2007. La Tension Narrative: Suspense, curiosité et surprise.
Paris: Seuil.
Barthes, Roland. 1966. “Introduction à l’analyse structurale des récits.”
Communications 8: 1–27.
Beauvisage, Thomas. 2001. “Exploiter des données morphosyntaxiques pour
l’étude statistique des genres: application au roman policier.” TAL 43.
http://www.revue-texto.net/Inedits/Beauvisage/index.html.
Biber, Douglas. 1988. Variation Across Speech and Writing. Cambridge:
Cambridge University Press.
Biber, Douglas. 2009. “A Corpus-Driven Approach to Formulaic Language:
Multi-word Patterns in Speech and Writing.” International Journal of Corpus
Linguistics 14: 275–311.
Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad, and Edward
Finegan. 1999. Longman Grammar of Spoken and Written English. London:
Longman.
Boyer, Alain-Michel. 2008. Les Paralittératures. Paris: Armand Colin.
12
Brunet, Etienne. 1981. Le Vocabulaire français de 1789 à nos jours. Geneva and
Paris: Slatkine and Champion.
Burton, Deirdre, and Roland Carter. 2006. “Literature and the Language of
Literature.” In Encyclopedia of Language and Linguistics, 2nd ed., Vol. 7,
edited by Keith Brown, Anne H. Anderson, Laurie Bauer, Margie Berns,
Graeme Hirst, and Jim Miller, 267–74. Boston: Elsevier.
Conrad, Susan, and Douglas Biber, editors. 2001. Variation in English: Multi-
dimensional Studies. London: Longman.
Coseriu, Eugenio. 1975. Sprachtheorie und allgemeine Sprachwissenschaft.
München: Fink.
Croft, William. 2001. Radical Construction Grammar. Oxford: Oxford
University Press.
de Beaugrande, Robert. 2005. “‘Corporate Bridges’ Twixt Text and Language:
Twenty Arguments Against Corpus Research and Why They’re a Right Load
of Old Codswallop.” http://www.beaugrande.com/Corporate%20Bridges.
htm.
Duff, David. 2000. Modern Genre Theory. London: Longman.
Eisenzweig, Uri. 1983. Autopsies du roman policier. Paris: Union générale
d’éditions.
Feilke, Helmuth. 1996. Sprache als soziale Gestalt: Ausdruck, Prägung und die
Ordnung der sprachlichen Typik. Frankfurt am Main: Suhrkamp.
Fillmore, Charles, Paul Kay, and Catherine O’Connor. 1988. “Regularity
and Idiomaticity in Grammatical Constructions: The Case of Let Alone.”
Language 64: 501–38.
Fischer-Starcke, Bettina. 2010. Corpus Linguistics in Literary Analysis: Jane
Austen and Her Contemporaries. London: Continuum.
Frow, John. 2005. Genre. London and New York: Routledge.
Goldberg, Adele. 1995. Constructions: A Construction Grammar Approach to
Argument Constructions. Chicago: The University of Chicago Press.
Gonon, Laetitia. 2012. Le Fait divers criminel dans la presse quotidienne
française du XIXe siècle. Paris: PSN.
Gonon, Laetitia. 2015. “Stylistique du vengeur dans les mystères de Paris
et quelques-uns de ses avatars, 1842–1847 [également disponible en
anglais].” Médias 19, Poétique des Mystères, Publications. Les Mystères
Urbains au XIXe Siècle: Circulations, Transferts, Appropriations, edited
by Dominique Kalifa and Marie-Ève Thérenty. http://www.medias19.org/
index.php?id=17807.
Gymnich, Marion, Birgit Neumann, and Ansgar Nünning, editors. 2007.
Gattungstheorie und Gattungsgeschichte. Trier: WVT.
13
Hausmann, Franz Joseph. 1979. “Un dictionnaire des collocations est-il pos-
sible?” Travaux de littérature et de linguistique de l’université de Strasbourg 17
(1): 187–95.
Hausmann, Franz Joseph. 2007. Collocations, phraséologie, lexicographie: études
1977–2007 et bibliographie. Aachen: Shaker.
Herbst, Thomas, and Michael Klotz. 2003. Lexikographie: Eine Einführung.
Paderborn: Schöningh.
Herrmann, J. Berenike, Christof Schöch, and Karina van Dalen-Oskam. 2015.
“Revisiting Style, a Key Concept in Literary Studies.” Journal of Literary
Theory 9 (1): 25–52.
Hoey, Michael. 2005. Lexical Priming: A New Theory of Words and Language.
London and New York: Routledge.
Hunston, Susan, and Gill Francis. 2000. Pattern Grammar. Amsterdam:
Benjamins.
Jakobson, Roman. 1921. Новейшая русская поэзия. Набросок первый. Prague:
Politika.
Kilgarriff, Adam, Vít Baisa, Jan Bušta, Miloš Jakubíček, Vojtěch Kovář, Jan
Michelfeit, Pavel Rychlý, and Vít Suchomel. 2014. “The Sketch Engine: Ten
Years on.” Lexicography 1 (1): 7–36.
Kraif, Olivier. 2016. “Le Lexicoscope: un outil d’extraction des séquences
phraséologiques basé sur des corpus arborés.” Cahiers de lexicologie 108:
91–106.
Lafon, Pierre. 1980. “Sur la variabilité de la fréquence des formes dans un cor-
pus.” Mots 1: 127–65.
Lebart, Ludovic, André Salem, and Lisette Berry. 1998. Exploring Textual Data.
Dordrecht: Kluwer.
Leech, Geoffrey. 2014. The Pragmatics of Politeness. Oxford and New York:
Oxford University Press.
Leech, Geoffrey, and Mick Short. 2007. Style in Fiction: A Linguistic
Introduction to English Fictional Prose, 2nd ed. London: Pearson.
Legallois, Dominique. 2006. “Quand le texte signale sa structure: la fonction
textuelle des noms sous-spécifiés.” Corela 5. https://journals.openedition.
org/corela/pdf/1465.
Legallois, Dominique. 2012. “La Colligation: autre nom de la collocation
grammaticale ou autre logique de la relation mutuelle entre syntaxe et
sémantique?” Corpus 11. http://corpus.revues.org/2202.
Legallois, Dominique, and Agnès Tutin. 2013. “Présentation: vers une exten-
sion du domaine de la phraséologie.” Langages 189: 3–25.
14
Lits, Marc. 2011. Le Roman policier dans tous ses états: d’Arsène Lupin à
Navarro. Limoges: Pulim.
Mahlberg, Michaela. 2013. Corpus Stylistics and Dickens’s Fiction. London and
New York: Routledge.
Mel’čuk, Igor, André Clas, and Alain Polguère. 1995. Introduction à la lexicolo-
gie explicative et combinatoire. Louvain-la-Neuve: Duculot.
Monte, Michèle, and Gilles Philippe, editors. 2014. Genres et textes: détermina-
tions, évolutions, confrontations. Lyon: PUL.
Mukařovský, Jan. 2014. “Standard Language and Poetic Language.” In
Chapters from the History of Czech Functional Linguistics, edited by Jan
Chovanec, 41–53. Brno: Masarykova univerzita.
Novakova, Iva. 2015. “Les Émotions entre lexique et discours.” In La
Sémantique et ses interfaces: actes du colloque 2013 de l’ASL, edited by
A. Rabatel, A. Ferara-Léturgie, and A. Létrugie, 181–204. Limoges:
Lambert-Lucas.
Novakova, Iva, and Elena Melnikova. 2013. “Vers un modèle fonctionnel pour
l’analyse du lexique des émotions dans cinq langues européenne.” Le bulle-
tin de la société de linguistique de Paris 108 (1): 131–60.
Nünning, Vera, and Ansgar Nünning. 2018. “Cultural Concerns, Literary
Developments, Critical Debates: Contextualizing the Dynamics of Generic
Change and Trajectories of the British Novel in the Twenty-First Century.”
In The British Novel in the Twenty-First Century: Cultural Concerns—Literary
Developments—Model Interpretations, edited by Vera Nünning, and Ansgar
Nünning, 21–52. Trier: WVT.
Quiniou, Solen, Peggy Cellier, Thierry Charnois, and Dominique Legallois.
2012. “Fouille de données pour la stylistique: l’exemple des motifs emer-
gents.” In Actes des 11es journées internationales d’analyse statistique des
données textuelles, Liège, 13–15 June 2012, 821–33. http://lexicometrica.
univ-paris3.fr/jadt/jadt2012/Communications/Quiniou,%20Solen%20
et%20al.%20-%20Fouille%20de%20donnees%20pour%20la%20stylis-
tique.pdf.
Rastier, François. 2011. La Mesure et le grain: sémantique de corpus. Paris:
Honoré Champion.
Renouf, Antoinette, and John Sinclair. 1991. “Collocational Frameworks in
English.” In English Corpus Linguistics: Studies in Honour of Jan Svartvik,
edited by Karin Aijmer, and Bengt Altenberg, 128–44. London: Longman.
15
Scholes, Robert. 1982. Semiotics and Interpretation. New Haven: Yale

University Press.
Scott, Mike, and Christopher Tribble. 2006. Textual Patterns: Key Words and
Corpus Analysis in Language Education. Amsterdam: Benjamins.
Siepmann, Dirk. 2005. “Collocation, Colligation and Encoding Dictionaries. Part
I: Lexicological Aspects.” International Journal of Lexicography 18 94): 409–44.
Siepmann, Dirk. 2011. “Sinclair Revisited: Beyond Idiom and Open Choice.”
In The Phraseological View of Language: A Tribute to John Sinclair, edited by Thomas
Herbst, Susen Faulhaber, and Peter Uhrig, 59–86. Berlin: Mouton de Gruyter.
Siepmann, Dirk. 2015. “A Corpus-Based Investigation into Key Words and
Key Patterns in Post-war Fiction.” Functions of Language 22 (3): 362–99.
Sinclair, John. 1991. Corpus, Concordance, Collocation. Oxford: Oxford
University Press.
Sinclair, John. 2004. Trust the Text: Language, Corpus and Discourse. London:
Routledge.
Stefanowitsch, Anatol, and Stefan Gries. 2003. “Collostructions: Investigating
the Interaction Between Words and Constructions.” International Journal of
Corpus Linguistics 8 (2): 209–43.
Steyer, Kathrin, and Annelen Brunner. 2014. “Contexts, Patterns, Interrelations:
New Ways of Presenting Multi-word Expressions.” In Proceedings of the 10th
Workshop on Multiword Expressions (MWE 2014), Gothenburg, Sweden, 26–27
April 2014, 82–88. https://aclweb.org/anthology/papers/W/W14/W14-0814/.
Stubbs, Michael. 2005. “Conrad in the Computer: Examples of Quantitative
Stylistic Methods.” Language and Literature 14 (1): 5–24.
Stubbs, Michael, and Isabel Barth. 2003. “Using Recurrent Phrases as Text-
Type Discriminators: A Quantitative Method and Some Findings.”
Functions of Language 10 (1): 61–104.
Tapanainen, Pasi, and Timo Järvinen. 1997. “A Non-projective Dependency Parser.”
In Proceedings of the Fifth Conference on Applied Natural Language Processing.
Association for Computational Linguistics. Washington, DC, 31 March–3 April,
64–71. http://delivery.acm.org/10.1145/980000/974568/p64-tapanainen.pdf.
Todorov, Tzvetan. 1980. Poétique de la prose: choix, suivi de nouvelles recherches
sur le récit. Paris: Éditions du Seuil.
Tognini-Bonelli, Elena. 2001. Corpus Linguistics at Work. Amsterdam and
Philadelphia: John Benjamins.
Zymner, Rüdiger. 2003. Gattungstheorie: Probleme und Positionen der
Literaturwissenschaft. Paderborn: Mentis.
2
The Notion of Motif Where Disciplines
Intersect: Folkloristics, Narrativity,
Bioinformatics, Automatic Text Processing
and Linguistics
Dominique Legallois and Stefan Koch
1 Introduction
In attempting to present and discuss the notion of motif in several
disciplinary fields, we have set ourselves an ambitious goal in this
chapter. While the distances that separate these disciplines represent a
challenge they also enrich the task before us.
So, why be interested in this notion of motif in the first place? It
seems to us that this notion is one of the few to transcend the bound-
aries between various areas of intellectual inquiry which otherwise may
have little in common. Then also, in recent years the notion of motif
has enjoyed unquestioned success in linguistics, specifically in the sub-
disciplines concerned with semantic or stylistic characterisation of texts.
D. Legallois (*)
University Sorbonne-Nouvelle, Paris, France
e-mail: dominique.legallois@sorbonne-nouvelle.fr
S. Koch
Friedrich-Alexander-University Erlangen-Nuremberg, Erlangen, Germany
e-mail: stefan.koch@fau.de
https://doi.org/10.1007/978-3-030-23744-8_2
18
D. Legallois and S. Koch
This success can be gauged by the proposals and innovations advanced

by various linguists in developing the concept or in creating tools that
automate the extraction of motifs from texts. The present volume in this
and the other chapters centers on applications of motifs.
The term motif is extremely complex: Even when tackled within the
same discipline and field of analysis, as in this volume, it is subject to
varied interpretations. To put it another way, the definition of motif
is still constantly evolving. This evolution is driven both by technical
advances in computer technology and conceptual innovations.
This dynamic compels us therefore to first equip linguists and literary
text specialists with the historical, epistemological and methodological
background on the use of motif in scientific research. Non-linguists, on
the other hand, may discover elements in the linguistic approach that
will let them reflect on and develop their own conception of motif. We
thus regard the term motif in its various conceptions as a potentially
solid basis for dialogue between subjects and disciplines that usually
work siloed from each other.
Our analysis begins with narratology and folkloristics, two fields in
which the notion of motif has been well represented, especially in nine-
teenth-century research on popular traditions, and has been further
developed since then by structuralists and narratologists/semioticians.
Next, we delve into the role the term motif plays in information
retrieval in science, both from the perspectives of Harris et al. (1989)
and of bioinformatics. As we will discover, these techniques for ana-
lyzing various pattern types are somewhat akin to certain of the older
methods in narratology.
A third set of approaches to motif is the one undertaken by Natural
Language Processing (NLP). Köhler (2006) had a seminal role in ini-
tiating recent work which developed the notion of motif as a discrete
linguistic unit that makes computational processing of texts and cor-
pora—characterization, clustering, classification, etc.—feasible.
Lastly, we turn to a series of linguistic studies carried out in France
during the past decade, including in the domain of computational lin-
guistics but also designed to cover several dimensions related to sty-
lometry, semantic characterizations, genre characterizations, etc. This
body of research was generated to a great extent in the course of the
2 The Notion of Motif Where Disciplines Intersect: Folkloristics …
19
PhraseoRom project on which all of the contributions in this book are

based.
At the end of this overview, the reader will have been presented with
a set of designs from which to draw conclusions that should equip him
or her to add value in assessing the possible ramifications of the concept
of motif.
Although motif frequently also crops up in music and painting, that
is, in nonlinguistic or non-textual domains, we confine our scope to
addressing and reflecting on textual motifs. Hence, we close this chapter
by applying the notion specifically to literary texts.
2 Motif in Folkloristics and Narratology

In cultural anthropology and folkloristics, the term motif refers to
a recurrent element in the popular traditions of a cultural commu-
nity. Several motifs combine to form patterns, allowing researchers to
describe, analyze and interpret cultural differences. Thus, broadly speak-
ing, motifs can be applied to the field of literature—including oral lit-
erature, such as folk tales—to visual and textile arts and to music. In
the latter two, motif then is either, respectively, any repeated decorative
element or the smallest distinctive recurring musical line from a melody
or rhythm. This musical motif is different from the one proposed by
Boroda (1982), which we discuss in more detail below.
2.1 Tales and Popular Narratives
For our purpose, the study of motifs in tales or popular narratives is of

particular interest. Here, motifs are recurrent microstories, recognizable
thanks to a stable but malleable and adaptable physiognomy, depend-
ing on the texts in which they occur. More precisely, they are stereo-
typical narrative sequences whose identity is definable and irreducible
(Vincensini 2000, 2), as shown by the following examples, some of
which are taken from Vincensini’s work:
20
(1) - “Punishment, sitting in the water”1, which appears in many novels

of the Middle Ages but also in the Lives of the Saints.
- “The perilous cemetery”: The hero is confronted at night by supernatu-
ral opponents (Cadot 1980).
- “The constrained gift”: A donor makes unlimited offers without even
being asked by the recipient. This motif appears in the Gospels of
Saint Matthew as well as in the Arthurian legend.
- “The accusatory bleeding”: A corpse starts bleeding in the presence
of the murderer. This motif is used in many medieval texts but also
appears in Le Second livre des Amours by Ronsard, in Richard III by
Shakespeare, in Les Tragiques by Agripa d’Aubigné.
- “The animal skin delimiting a territory”: This motif concerns a trick
used by the hero to claim a large territory.
The conception of motif in ancient and traditional texts necessarily

touches on intertextuality and collective memory. In a sense, motifs are
always leitmotivs in a culture.
Based on the first collections made by the Grimm brothers, in the
Verzeichnis der Märchentypen, published in 1910 by the Finnish folklor-
ist Antti Aarne (1867–1925; see Aarne [1961] for an English version),
taletypes (Märchentypen ) are classified into categories, each assigned a
number and a title in German. The concept of the taletype accommo-
dates variations of the same tale in different countries or repertoires. The
American folklorist Stith Thompson (for example 1946) continued this
work, which ended up in the international classification of tales known
as the Aarne-Thompson classification. But he also proposed a catego-
rization of motifs that resulted in the famous Index of Motifs found in
the six-volume Motif-index of folk-literature: a classification of narrative
elements in folktales, ballads, myths, fables, mediaeval romances, exempla,
fabliaux, jest-books, and local legends, published between 1932 and
1936.2 It sets up the following main categories:
1This motif is also what was called an exemplum (pl. exempla), that is, a short moral narrative
given as true and used for educational purposes in sermons in the twelfth and thirteenth centu-
ries. exempla were also used to “christianize” traditional stories. There were exempla repertories,
for example, the scala coedi collection by Jean Gobi dated around 1300 (Vincensini 2000, 13).
2A revised and enlarged edition was published between 1955 and 1958.
21
a) mythological motifs,
b) animals,
c) tabu,
d) magic,
e) the dead,
f) marvels,
g) ogres,
h) tests,
j) the wise and the foolish,
k) deceptions,
l) reversal of fortune,
m) ordaining the future,
n) chance and fate,
p) society,
q) rewards and punishment,
r) captives and fugitives,
s) unusual cruelty,
t) sex,
u) the nature of life,
v) religion,
w) traits of character,
x) humour, and
z) miscellaneous groups of motifs.
Thompson divided each category into subcategories. For example, the

motifs related to the wise and the foolish he broke down into the follow-
ing subcategories:
- J0-J199 Acquisition and possession of wisdom/knowledge

- J200-J1099 Wise and unwise conduct
- J1100-J1699 Cleverness
- J1700-J2799 Fools (and other unwise persons)
In his book The Folktale (1946), Thompson described the many distinc-
tive classes as follows:
22
A motif is the smallest element in a tale having a power to persist in tra-

dition. In order to have this power it must have something unusual and
striking about it. Most motifs fall into three classes. First are the actors
in a tale – gods, or unusual animals, or marvelous creatures like witches,
ogres, or fairies, or even conventionalized human characters like the
favorite youngest child or the cruel step-mother. Second come certain
items in the background of the action – magic objects, unusual customs,
strange beliefs, and the like. In the third place there are single incidents
– and these comprise the great majority of motifs. (Thompson 1946,
415–16)
A current conception of motif in narratology seems to echo Thompson’s

comments. The striking character of the motif, so to speak its expressive
character, lies in its concrete value. In a recent reference book on narra-
tology, Abbott (2008, 95) after Prince (2003, 55) states, that “[m]otifs
are, in Gerald Prince’s words, the ‘minimal thematic unit’ (Dictionary,
55).” Abbott then goes on to contrast motif and theme (see Abbott
2008, 237, 242) as follows:
Motif: A discrete thing, image, or phrase that is repeated in a narrative.

Theme, by contrast, is a more generalized or abstract concept that is sug-
gested by, among other things, motifs. A coin can be a motif, greed is a
theme. (Abbott 2008, 237)
Theme: A subject (issue, question) that recurs in a narrative through
implicit or explicit reference. Along with motif, theme is one of the two
commonest forms of narrative repetition. Where motifs tend to be con-
crete, themes are abstract. (Abott 2008, 242)
Abbott (2008, 95) also illustrated his definitions with the following
example: “Windows, for example, are a motif in Wuthering Heights and,
given the way Brontë deploys them, they support a highly complex
interplay of three themes: escape, exclusion, and imprisonment. When,
for another example, the character Barkis in David Copperfield continues
to repeat his cryptic phrase, ‘Barkis is willin’,’ it becomes a motif, a sig-
nature phrase for the theme of shy, honest-hearted devotion in love that
23
Barkis exhibits in his pursuit of Peggotty.” The motif in contemporary

narratology is therefore a minimal unit that “concretizes” the themes.3
2.2 The Russian Formalists
Since the 1920s, the Russian formalists have also regarded the motif as
a minimum unit but have also pointed out the difficulty of defining this
minimality. Additionally, these researchers developed elements for artic-
ulating the relationship between motifs and linguistic forms.
Thus, Tomaševskij (1925 [1965]) equated motif and clause (see also
Ducrot and Todorov 1972). To our knowledge, this was the first time
anyone reflected on the linguistic dimension of motif. We will elabo-
rate on this point below. Tomaševskij (1925 [1965]) also proposes that
some motifs can be dynamic in how they change the narrative situation
or make it evolve. Motifs that do not change the narrative situation are
termed stative motifs. We recognize this also as a first reflection on the
gestalt complementarity between foreground and background or figure
and ground. Tomaševskij also famously distinguished fabula from story,
sometimes referred to as the story/narrative distinction.
For Veselóvskij (1913a, b), each proposal had its own motif which
he held to be the smallest unit of the thematic material. The notion of
motif thus was radically redefined, further distancing it from the folk-
lorist conception. However, according to Propp (1968 [1928]), by pos-
iting non-compositionality for the linguistic clause, we lose sight of all
the possible substitutions that make up the richness and variety of tales:
If a motif is something logically whole, then each sentence of a tale

gives a motif. (A father has three sons: a motif; a stepdaughter leaves
home: a motif; Ivan fights with a dragon: a motif; and so on.) This
would not be so bad if motifs were really indivisible; an index of motifs
would then be made possible. But let us take the motif ‘a dragon kid-
naps the tsar’s daughter’ (this example is not Veselóvskij’s). This motif
3See for example the helmet motif in L. F. Céline’s Casse-Pipe, studied by Richard (1979).
24
decomposes into four elements, each of which, in its own right, can
vary. The dragon may be replaced by Koščéj, a whirlwind, a devil, a
falcon, or a sorcerer. Abduction can be replaced by vampirism or vari-
ous other acts by which disappearance is effected in tales. The daughter
may be replaced by a sister, a bride, a wife, or a mother. The tsar can be
replaced by a tsar’s son, a peasant, or a priest. In this way, contrary to
Veselóvskij, we must affirm that a motif is not monomial or indivisible.
(Propp 1968, 13)
Propp sought to replace Veselóvskij’s (1913b, 9–12) motif with the

notion of narrative function, that is, an action defined in terms of its
meaning in plot development independent of the characters and how
these functions are fulfilled. Following the French philologist Bédier,
Propp (1968) differentiates constant patterns from the variable patterns,
calling the former function and the latter attribute. The names of the
characters may change but their actions or functions do not.
The 31 narrative functions in the traditional Russian tale are certainly
motifs—narrative atoms—but Propp prefers to dispense with a cumber-
some terminology. The functions (for example: acceptance of the mission
by the hero; departure of the hero; testing of the hero by a donor, etc.) find
their place in sequences—they are ordered and arranged, as in these
examples: lack—compensation for lack, prohibition—violation of the pro-
hibition, combat—victory, and so on.
2.3 Some Developments of the Formalist Approach
Propp’s model made an impact from the 1960s onwards. Thus, the
American folklorist Alan Dundes (1934–2005) used Pike’s linguistic
terminology (1967) in an attempt to apply Propp’s method to a corpus
of Amerindian stories. Dundes (1962) proposed renaming the Proppian
function motifemes. He also developed the term allomotif to designate
the various forms in which the motif is realized in the tale. Allomotifs
are to motifemes as allophones are to phonemes.
The Czech researcher Doležel (1972) repatriated the notion of
motif to the field of literary text analysis. He approached it on three
levels: motifeme, structural motif and texture motif. The first level is
25
metalinguistic. It consists of assigning a series of expressions designating

an action to another series of expressions defining the actant. Doležel
in effect defines the motifeme as a clause similar to Propp’s function. The
second level, also metalinguistic, focuses on the narrative segments by
which an action is assigned to a character. Indeed, in Doležel’s concep-
tion the structural motif is limited to dynamic motifs only. Finally, at
the third level, the texture pattern corresponds to a sentence in the text
itself. In the author’s words:
Examples of motifeme strings; (a) one-actant strings: The hero returned;

the hero passed the test. (b) Two-actant string: The hero defeated the
villain.
A MOTIF (m) is a proposition predicating an action to a character.
Examples of motif strings: Ivan killed the dragon; Yanko rescued from
death a grandchild of Swaffer; the judge sentenced Meursault to death.
A MOTIF TEXTURE is a narrative sentence (i.e. a sentence of the narra-
tive text).
(2) And suddenly, one fine morning in spring, he rescued from an
untimely death a grandchild of old Swaffer (Conrad).
Le president m’a dit dans une forme bizarre que j’aurais la tête tranchée sur
une place publique au nom du peuple francais (Camus) (The president told
me in a strange way that my head would be cut off in a public square on
behalf of the French people).
The motifeme/motif relationship will be called SPECIFICATION. We
will say that a motifeme is specified by a motif (motif sequence) or that a
motif (motif sequence) specifies a motifeme. The motif/texture relation-
ship will be called VERBALIZATION. A motif will be said to be ver-
balized by a texture; a texture will be said to verbalize a motif. (Doležel
1972, 59–61; translation Camus DL)
These perspectives reflect a need to conceive of motif on several lev-

els, including an abstract level (for example the motifeme ) of a stereo-
typed entity identifiable in a literary culture or tradition and which is
linguistically embodied in various forms (for example the texture
motif).
To round up this—necessarily incomplete—narratology survey, we
need to address a highly formal aspect of motif, which manifests itself
26
especially in Propp’s criticism of Vólkov. Vólkov regards a motif as a

quality of the heroes (e.g. |two wise sons-in-law and the third a madman|),
an act by the heroes (e.g. |the injunction of a father for someone to watch
over his grave after his death, an injunction that is executed by the madman
alone|), an object (e.g. |houndstooths|, |talismans|). In Vólkov’s annotation
system, each motif is given a conventional sign consisting of a letter and
one or two numbers. Similar motifs carry the same letter but different
numbers. This formalization allows Vólkov to translate the motifs into
alphanumeric signs and thus to compare combinations. This method
seems naive to Propp (1968, 15): “The only ‘conclusion’ that can be
drawn from this transcription is that similar tales resemble each other – a
conclusion which is completely noncommittal and leads nowhere.” What
interests us is not the conception of motif in Vólkov’s work, but rather his
formalization process, which is strikingly analogous to recent research not
into narratives or tales but into a completely different field, that of bioin-
formatics. How could bioinformatics and narratology possibly be related?
The motifs of Russian or Amerindian tales after all take root in stories
that are very different from those told by genome sequences. But just
maybe they have something to say to each other regarding methodology.
3 From Narratology to Bioinformatics

Using the notion of sublanguage, Harris et al. (1989) showed how
scientific reports can be “translated” through a set of formal transfor-
mations into a sequence of formulas representing the information con-
tained in the reports. It primarily involves forming classes of words that
relate grammatically to specific other words. For example, words of class
A may appear as subject of the phrase found in the lymph nodes after
injection of an antigen:
These “A” words could include antibodies, agglutinin, etc. With this as
starting point, we can then form a class V of operators, such as is found
in, is contained in, is produced by, that take A words as their subjects (e.g.
antibodies, agglutinin ). Proceeding in this way, we find that specific word
classes recur in a particular grammatical relation to certain other word
27
classes, e.g. A words as subjects of V operators with T objects or “comple-

ments” (lymph nodes, lymph, serum ). Together, they create a sentence type
(structure) AVT. It is simultaneously generated with the process of setting
up (by extension) the word classes A, V, and T, because these are defined
from the start by their concatenation in the operator-argument, i.e., a sen-
tence structure. The grammatical transformations applied earlier will have
reinforced this result, since, for example, the sentence lymph nodes produce
antibodies will have been recognized as a transform of antibodies are pro-
duced by lymph nodes, hence as an instance of AVT. (Harris et al. 1989, 8)
This procedure could be computerized to identify recurrent, informa-

tionally relevant patterns in scientific papers. Following Darányi et al.
(2012), comparing it with the morphology of tales holds considerable
interest for us: although tales have been digitalized, there is nothing in
the literature to indicate they have been subjected to this sort of precise
computational analysis. We believe it is feasible to analyze combinations
of motifs from a very large data set of occurrences. The technique con-
ceivably could emulate how Harris et al. used sublanguages or methods
used for motif detection in infobiology to detect relationships between
genes and rare diseases in biomedical texts for example.
Since Frederick Sanger started DNA sequencing in the late 1970s,
the technology has advanced by leaps and bounds, resulting in exponen-
tial increases in the volume of available DNA sequences. Infobiologists
call the recurring sequences they detect by aligning sequences (analo-
gous to the linguist’s concordances ) motifs that let them find similarities
between two sequences and determine their possible homologies. For
example, in the genome sequences in Fig. 1 a special pattern emerges
(based on D’haeseleer 2006, 423).
This domain-specific practice from the life sciences finds an analogue
in formal descriptions of narrative content. Darányi et al. (2012) there-
fore proposed to deal with narrative motifs in the same way as motifs
from biomedical texts: patterns extracted from a set of tales indexed in
The Aarne-Thompson-Uther Tale Type Catalog receive an alphanumeric
annotation, thus reducing each tale to a set of annotations such as the
one in Table 1 for the The Dragon-Slayer tale and its variants showing a
closed motif structure.
28
+(0 &&&$77*77&7&
+(0 777&7**77&7&
+(0 7&$$77*777$*
$1% &7&$77*77*7&
$1% 7&&$77*77&7&
$1% &&7$77*77&7&
$1% 7&&$77*77&*7
52; &&$$77*7777*
Fig. 1 Genome sequences
Table 1 The Dragon-Slayer as a motif chain and its equally valid story variants
(based on Ofek et al. 2013, 3)
$787DOH7\SH7KH'UDJRQ VOD\HU
% % % 6 7 ' ' % % 5 + .
6HTXHQFH 9DULDQWV
% % % 7 ' ' % + . . + +
% % 6 7 ' ' % + . . + +
% % % 7 ' ' % + . . + +
% % 6 7 ' ' % + . . + +
% % % 7 ' ' 5 + . . + +
% % 6 7 ' ' 5 + . . + +
% % % 7 ' ' % + . . + +
% % 6 7 ' ' % + . . + +
% % % 7 ' ' % + . . + +
% % 6 7 ' ' % + . . + +
% % % 7 ' ' 5 + . . + +
% % 6 7 ' ' 5 + . . + +
% % % 7 ' ' % + . . + +
% % 6 7 ' ' % + . . + +
% % % 7 ' ' % + . . + +
% % 6 7 ' ' % + . . + +
% % % 7 ' ' 5 + . . + +
% % 6 7 ' ' 5 + . . + +
% % % 7 ' ' % + . . + +
% % 6 7 ' ' % + . . + +
% % % 7 ' ' % + . . + +
DNA motif sequences and narrative motif sequences are identical in

several respects, including some transformations:
Adding grammar and moving over to permutations, sequences start

to play a role. Canonical nucleotide sequences generate secondary and
tertiary – in fact spatial – structures such as the famed double helix;
canonical motif sequences may contribute to the evolution of tale types,
themselves representatives of tale variants. Whereas most mutations are
29
neutral and have little or no impact on the functionality of the product,

their adding up can dramatically affect the survival rate of the outcome,
leading to new genotypes and phenotypes in the course of evolution. In
the same vein, deletion and translocation could be standard tools in the
narrative building toolkit; inversion is suggested to play a central role
in the Bible (Christensen, 2003), and duplication is evident, e.g. in the
case of the Proppian narrative scheme where complete tale moves may
be repeated several times or combined with one another by different
embeddings (Propp, 1968). This indicates the need for a theory of text
evolution as a series of narrative element recombinations, forming from
simple to more complex structures by “mutation mechanisms”. (Darányi
et al. 2012, 4)
These similarities in motif analysis in two very different scientific

fields show the great collaborative potential that exists between natural
scientists, computer scientists, narratologists and linguists.
4 Motifs in Automatic Text Processing

Most recently, the term motif for use in automatic text processing has
been defined especially on Köhler’s initiative (first in 2006; see also
2015). In his definition, a motif is the longest continuous sequence of
equal or increasing values representing a linguistic unit’s quantitative
property. Considering the motif as a unit that can reflect the sequen-
tial organization of a text without resorting to grammatical or linguis-
tic analysis is the key here. Without going at length into the reasons
why, let us present an example taken from Köhler (2015, 92) in which
the sentence “word length studies are almost exclusively devoted to
the problem of distributions” is analyzed in light of the above defini-
tion of motif as the “longest continuous sequence of equal or increasing
values…”
Counting the syllables in each word (word = 1, length = 1, stud-
ies = 2 and so on), yields a sequence of 5 motifs, so-called word length
motifs: (1-1-2) (1-2-4) (3) (1-1-2) (1-4). A word with fewer sylla-
bles than the previous word becomes the first element of a new motif.
Applied to an entire text, the method according to Köhler provides the
30
insight that motifs display a rank-frequency distribution of the Zipf-

Mandelbrot type. As such, they behave like traditional units of linguis-
tic analysis. Hence, the method makes comparing texts and authors
or performing textual classification possible. However, the example of
“syllabic” patterns is not the only one: motifs can also be defined for
any linguistic unit (phoneme, sentence type, type of clause, etc.). Here,
Köhler (2015, 98) once again furnishes an example in a more semantic
vein, that is, the segmentation of newspaper commentaries into argu-
mentative relationships—[“elaboration”], [“elaboration,” “concession”],
[“elaboration,” “evidence,” “list,” “preparation,” “evaluation,” “conces-
sion”], [“evidence,” “elaboration,” “evaluation”]. Again, these motifs
serve as the basis for establishing textual typologies.
A final example comes from Čech et al. (2017) who take verbal
valence as elements of the motif. To each verb in a text they assign a
number corresponding to the number of its arguments, which results in
an array like the one shown in Fig. 2.
The motifs and their frequency are then identified. Shown here are
the first five in the upmost row (Fig. 3).
>

@
Fig. 2 Verbal valence as elements of the motif (based on Čech et al. 2017, 16)

DQGVRRQ
Fig. 3 Frequency of the motifs (based on Čech et al. 2017, 16)

31
Once again, the study shows that this motif type can be considered
as a unit sharing the same distributional patterns as well-established
traditional units, including the rank frequency and spectrum abide by
the Zipf-Mandelbrot distribution. The authors emphasize that valency
motifs can be considered regular language entities.
Interestingly, Köhler based his approach and the design of “his”
motif on the so-called F-motif conceived by the musicologist
Boroda in the 1970s and 1980s (Boroda 1973, 1982) as a unit in
music corresponding to a word in language. His F-motifs, just like
Köhler’s linguistic motifs, are sequences of equal or increasing val-
ues, only that Köhler’s syllables are Boroda’s tones. Boroda had also
tested the distributional patterns of his motifs against the Zipf-
Mandelbrot law only to find that they matched nearly perfectly the
distribution projected by it. Köhler imported Boroda’s concept—
which, it should be noted, does not correspond to the classic notion
of motif in musicology as a recurring melody, etc. (see above)—to
linguistics.
While this conception of motif as a unit of textual analysis is still
limited to the formal level, this does not prevent its being extended to
more semantic considerations, as the study on argumentative relation-
ships makes clear. We will let Köhler (2015, 107) have the last word
on this:
Motifs provide a means to analyse texts for their sequential structure with
respect to all kinds of linguistic units and properties; even categorical
properties can be studied in this way. The granularity of an investigation
can be adjusted by iterative application of motif-formation, and proven
statistical methods can be used for the evaluation. The full potential of
this approach has not yet been explored.
This method of automatic text processing proposed by Köhler and col-

leagues differs profoundly from the “bag-of-words” model prevalent
in linguistics and stylistics, whose flaw lies in failing to deal with the
sequentiality of the units composing a text or their topology.
32
5 The Contribution of Computer Linguistics

in the French Domain
In recent years, several approaches have been developed in France, first
independently and then collaboratively in part, that combine elements
of both computer processing and text linguistics. The main objective was
to create methods and tools for identifying what in these approaches are
also called motifs. In the following, we provide an overview of these per-
spectives, all of which view motif as a statistically significant unit or a
unit composed of statistically significant linguistic representations of it.
5.1 The Ganascia (2001) Approach
Ganascia advanced his approach to textual motifs in one of the first

works on the subject that he published in France in 2001. In this case,
it is opportune to speak of textual motifs because it is indeed a question
of characterizing speech. However, these characterizations are of a syn-
tactic nature—in other words, texts are characterized by the syntactic
specificities of the sentences that compose them.
Ganascia automatically extracts recurrent sequences from syntactic
trees. Applied to texts, his method, as implemented, makes it possible
to identify their stylistic characteristics. More recently, Ganascia and his
team at the Laboratoire d’Informatique de l’université Paris 6 have been
working on a simpler model for identifying motifs of POS in texts by
factor analysis. The text is first segmented into a set of sentences and
then each sentence is mapped into a sequence of syntactic (PoS tag)
items, as in the following sentence4:
(3) “J’aime ma maison où j’ai grandi.”

(I love my house where I grew up)
This is first mapped onto a sequence of PoS tags:
<PRO:PER VER:pres DET:POS NOM PRO:REL PRO:PER
VER:pres VER:pper SENT>
4From Frontini et al. (2018, 122).

33
In Frontini et al. (2018), the authors characterize the style of the

dialogues between characters in Molière’s plays. Thus, the two comical
protagonists Scapin and Harpagon are each characterized by patterns of
lower syntactic complexity. This is especially the case with Harpagon,
whose patterns convey the image of a self-centered person insistent on
getting his way but frustrated by violent disappointments (especially
when it comes to money):
(4) - [PRO:PER] [PRO:PER] [VER:pres] [VER:pper]

- on m’ a privé … [they have deprived me …]
- on m’ a dérobé … [they have robbed me …]
- on m’ a volée … [they have stolen from me …]
- on m’ a pris … [they have taken me …]
The technique is partly implemented in the Eremos online platform

(http://eremos.lip6.fr/index.php). Eremos is a web application capa-
ble of extracting and manipulating syntactic patterns via a simple, fast,
user-friendly interface.
Another platform called SDMC (https://tal.lipn.univ-paris13.fr/
sdmc/; see Sahraoui et al. 2017) which is also dedicated to the iden-
tification of syntactic patterns deserves mention here: it extracts reg-
ularities from a text or a corpus, that is, sequences of words, lemmas
and POS tags that frequently appear together. A set of filters can be
applied to constrain the extracted patterns: minimum frequency, or
text-specific constraints like “contains an adjective, or a verb.” SDMC
facilitates the analysis of these sequential motifs with a visualization.
SDMC makes it possible to search not only for one motif but also for
their co-occurrences. On the left side of the visualization appears the
index of motifs while in the middle column we see the sentences con-
taining the motif. On the right side of the screen, the tool proposes
a compact version of the text with the sentences replaced by stars. In
this way, SDMC facilitates analyzing the topological dimension of a
pattern.
34
5.2 The Approach of Longrée and Mellet
Longrée and Mellet (2018, 156) argue for considering motif as a textual
unit from the perspective of the textometry of Latin texts:
What is a motif? In a formal way the motif is defined as an ordered sub-

set of the textual ensemble, formed by the recurring combination of n ele-
ments provided with its linear structure. Thus, if the text is formed by a
certain number of occurrences of elements A, B, C, D and E, a motif can
be the recurring microstructure ACD or AAA, etc., without here prejudg-
ing the nature (lexical, grammatical, metrical, …) of the elements A, B, C,
D and E in question: the ‘motif ’ is only the framework – or the colloca-
tional pattern – accommodating a range of parameters to be defined and
capable of characterizing the diverse texts of a corpus or even the different
parts of a text. The motif is involved in the temporal dynamics of the nar-
ration, in the relations between sentences, and between the different tex-
tual sequences, such as descriptions, narrations, argumentations and so on.
As general pattern, the motif is able to characterize a genre; but its different
realizations or tokens may be specific to different authors in a given genre.
The objective here is clearly the topological characterization of texts.

The authors understand text as an ordered and multidimensional space
and aim to highlight the units that contribute to organizing this space.
Their definition emphasizes the nonlinear dimension of the pattern and
its variability, as the following example5 shows:
(5) here I would like/wish to specify however

here, however, I would like to clarify
Ø I wish to specify, however, that
here I would like to specify Ø
here I would like to specify however
The authors’ view currently tends toward a cognitive validation of

motifs. It involves linking the associations between the lexical elements
5Our example.
35
of a motif as measured by textual statistics and the priming effects of

psycholinguistic analysis.
5.3 The RLTs Approach
Also investigating the notion of motif is the PhraseoRom project for

identifying lexical pivot patterns specific to six novel subgenres (crime,
romance, historical, science fiction, fantasy and general fiction). The
project participants use a method based on corpora of syntactic depend-
ency trees involving the automatic extraction of recurrent lexico-
syntactic trees (RLTs) (see Kraif 2016). RLTs are lexical units linked by
syntactic dependency relationships. They are constructed from statisti-
cally significant series of co-occurrences based on a statistical associa-
tion measure, which in this context means calculating the log-likelihood
ratio (LLR) using the keywords method (Bertels and Speelman 2013).
The Lexicoscope program (Kraif 2016) is the automatic extraction tool
used. RLTs extraction starts with a lexical pivot (verbal or nominal)
selected by frequency and dispersion criteria from among the authors
composing the corpus and ends with producing a set of recurrent
subtrees.
The project uses the following criteria to extract RLTs specific to each
of the novel subgenres in question:
– A frequency threshold for nominal and verbal pivots relative to the

sub-corpus size (greater than 5 or 10);
– An RLTs size of at least 3 elements, with no upper limit;
– An LLR greater than 10.83;
– A variable dispersion according to the compared corpora, for example
up to 50% of the authors making up the corpora.
Figure 4 shows a sample extracted RLT corresponding to the expres-

sion il pousse la porte (he pushes the door open), specific to French
crime novels when compared to contemporary French general literature
(Gonon et al. 2018).
36
Fig. 4 An example of the RLT pousser la porte (push the door open)
Compared to purely sequential approaches, such as n-grams, RLTs

extraction has the advantage of associating only elements that are
united by a syntactic relationship, which considerably reduces noise.
Furthermore, RLTs make it possible to free oneself from linearity
by exploiting syntactic relationships and thus limiting redundancy.
Examples (6) and (7) correspond to a single RLT in which cœur (heart)
is the subject of the verb battre (to beat).
(6) Le coeur de Kurt battait à coups immenses. (Berthelot Khanaor

1 – Solstice de fer, 1985)
(Kurt’s heart was beating with immense blows).6
(7) Le coeur encore battant, il se rendit compte que c’était lui qui avait
crié. (Achard Tout seuls, 2012)
(His heart still beating, he realized that it was he who had screamed).
These two examples also illustrate another special feature of RLTs: they
allow gaps between the elements forming a pattern, which reduces the
redundancy that inevitably accompanies a sequential approach. Other
steps are thus still feasible. Indeed, since this method is based on the
lexicon, as many RLTs can be extracted as there are paradigmatic and
syntagmatic variations of the motif. For example, just like the RLT
6All of the following translations are our suggestions, unless otherwise specified.
37
Table 2 Textual motif “discovery of a body” (based on Muryn et al. 2016, 9–11)
N <HUM> <DESC> <POSITION> <MANNER> <LOC>
The corpse of Disarticulated Huddled up ∅ In the middle of a
a little girl Naked Lying Unconscious forest
A man In a house
in the 13th
arrondissement
“open the door” presented above, the RLTs “close the door,” “push the
door,” “open the door” and “the door opens” are specific to crime fic-
tion (compared to general literature fiction). A script to automatically
group these RLTs based on a vectorial analysis of their similarity is
under development in the PhraseoRom project. From these groupings,
we can then study the discursive functions of motifs (see Chapter 1 by
Novakova and Siepmann, Section 3, in this volume and Appendix B).
In other words, the motif is not only automatically identified, it is also
subjected to the analyst’s interpretation.
Evidently, this conception of motif could be linked to more narratolog-
ical perspectives, such as the one proposed by Muryn et al. (2016) in their
work with the notion of semantic motif in crime fiction: “The semantic
motif is an abstract representation of all structures realizing the same predi-
cate-argument model … (it) can thus take the form of a complex sentence,
a simple sentence, an SN, etc., or be inferred in whole or in part” (Muryn
et al. 2016, 4). In our view, we are close here to the complex conception of
Doležel’s (1972) motif as presented above. Thus, one of the motifs inher-
ent in the crime scene in a crime fiction novel is the discovery of a body.
This narrative moment (motifeme) is expressed by various possibilities (tex-
ture motifs) but can be abstracted into a set of headings (Table 2).
In this manner, Muryn et al. (2016) propose to create a semantic
grammar for a particular literary genre.
5.4 Motifs and Constructions
Certain approaches based on n-grams or lexical bundles (for example,

Mahlberg 2012) aim to characterize, among other dimensions, how
personae in novels are constructed without resorting to the term motif.
38
This is no longer a matter of narratological motifs in the sense of

minimal concrete themes but is a type of fundamental research into
writing processes and to analyzing the specific features of textual genres.
The following example is from Siepmann (2015, 381), who is interested
in units that he considers to be lexically specific narrative patterns and
which he calls key word patterns or syngrams:
(8) PossDet + thoughts were on / of NP / V+ing
This pattern then introduces a passage depicting a character’s train of
thought:
Like the categories of discourse markers known as ‘announcers’ or ‘topic

initiators’ (Siepmann 2005:93), which prepare the reader for the next step
in the author’s exposition or reasoning (I will now briefly describe ) or ini-
tiate a new topic in the current text, this kind of unit frequently, though
not systematically, serves to indicate a change in focalization, most typi-
cally a shift from external to internal focalization. It is notably common
for this marker to be followed by a proper noun.7 (Siepmann 2015, 381)
The work carried out by Legallois (see Legallois et al. 2018) attempts
to identify these types of patterns but also ones that are formally more
abstract. We can thus speak of grammatical motifs, particularly with
respect to the method first presented by Quiniou et al. (2012) with which
the authors sought to identify recurring lexico-grammatical patterns in
a corpus of nineteenth-century poetry. Since then, the method has been
developed and applied to different corpora, especially in distinguishing
textual genres, for example: between travel stories and fiction by the same
authors (see also Diwersy and Legallois, forthcoming), the characteristics
of the Harlequin novels (see Legallois et al. 2016) or the stylistic motifs in
nineteenth-century French novelists (see Legallois et al. 2018).
The method is based on a hybrid annotation of corpora, in which
the set of labels corresponds to general morpho-syntactic categories for
7And, parting from them abruptly, he ran down the stairs and out into the air. He bolted into a
hansom, and drove to the Goat’s Club. His thoughts were on Holly and what he must do before
her brother showed her this thing in tomorrow’s paper (Galsworthy In Chancery, 1920).
39
nouns, adjectives and for most verbs and adverbs. The most common
verbs (auxiliaries, modals, etc.) are reduced to their lemma. The form
of invariable words (prepositions, conjunctions, certain adverbs, etc.) is
preserved. For example,
(9) “La lecture a ceci de magique qu’elle permet de remonter le temps”
(Thilliez L’Anneau de moebius, 2008) (Reading is magical in that it allows
us to go back in time)
is annotated as:
(10) le N avoir ceci de ADJ que_SUB il permettre de INF le N.
Then, by extracting repeated segments of variable length, we obtain a set
of sequential patterns. Finally, the specificity of each pattern for a given
text relative to the other texts in the corpus is statistically calculated—
the idea being to characterize these texts by units that are anything but
lexical. We demonstrate the type of unity that can be captured this way
by reverting to differentiating between the crime genre and romance
novels. The relatively small corpus here includes:
– 12 “detective” novels, written by Thilliez, Benacquista, Vargas and

others (2 texts per author), 479,129 total words
– 12 novels from the Harlequin collection, 336,902 total words
We are only interested here in a few motifs related to the thriller. For
us, these motifs have a construction value, in the sense of construction
grammar (for example, Croft and Cruse 2004; Goldberg 2006). A con-
struction is a linguistic unit (a sign), which can be a morpheme, a word,
the argument schema of a verb, or a phraseological unit. Constructions
can be generic (such as the transitive construction “subj V Obj”),
semi-specific (“subj break obj”) or specific (give the Devil his due ).
A construction is associated with a meaning that can be semantic, prag-
matic or functional.
The theoretical framework of construction grammar is an important
contribution to understanding the role of motifs in texts. Here, the
motif is a construction—a linguistic sign—considered not only from
the grammatical point of view but from the perspective of its discursive
40
dimension in a particular text or a particular genre. Nevertheless, these

motifs, like the constructions studied in construction grammar, contrib-
ute to the linguistic competence of “speakers” (here writers and readers),
a competence understood as knowing not forms as such but how to use
them.
Given the small size of the corpus, the number of occurrences of
these motifs naturally is modest. Nevertheless, their distinctive dimen-
sion is obvious. This is how we remember:
(11) Motif [Inf, inf, inf…] (8 occurrences)

(a) Mettre le feu dans le passé, tout perturber, créer des failles, des
incohérences, et, peut-être, permettre l’arrestation de l’assassin,
avant qu’il s’en prenne à Sylvie (Thilliez L’Anneau de moebius,
2008) (Set fire to the past, disrupt everything, create loopholes,
inconsistencies, and perhaps allow the murderer to be arrested
before he attacks Sylvie).
(b) Foncer à Méry-sur-Oise, faire le tour des établissements scolaires,
comprendre quel rôle elle jouait dans cet incroyable micmac (Go
to Méry-sur-Oise, visit all the schools, understand what role she
played in this incredible little game). (Thilliez L’Anneau de moe-
bius, 2008)
(c) Fuir, faire demi-tour, trouver le réseau et alerter les renforts (Run,
turn around, find the network and alert the reinforcements).
(Thilliez La Chambre des morts, 2005)
Here the motif is an accumulation of infinitives referring to important

trials in the narrative, a pattern we find essentially in all of Thilliez’s
work. The content of these adverbial sentences is anchored in the char-
acter’s inner monologue, which urgently produces a reflection to deter-
mine a course of action.
(12) Motif [N, N, N, …] (7 occ.)

(a) Catacombes, laboratoires de biologie, nécropoles, hôtels
bizarroïdes, et j’en fais ensuite une espèce de grosse soupe, dans
mon cerveau (Thilliez La Chambre des morts, 2005). (Catacombs,
biology laboratories, necropolises, weird hotels, and then I make
a kind of big soup out of it, in my brain)
41
(b) Depuis l’automne, mélancolie, mutisme, brusquerie, insomnies

et sautes d’humeur (Vargas Coule la Seine 2002). (Since autumn,
melancholy, silence, abruptness, insomnia and mood swings)
This time we have an accumulation of nouns (concrete in example (a),

abstract in (b)) that summarizes past dysphoric experiences. The motif
also gives us access to the consciousness of the character who is in a state
of complete confusion.
(13) Motif [pas de N] (9 occ.)

(a) En pleine nuit, pas de lumières, il n’avait aucune chance d’être
vu. (Thilliez La Chambre des morts, 2005) (In the middle of the
night, no lights, he had no chance of being seen.)
(b) A priori, pas de traquenard (Benacquista Malavita, 2002) (A pri-
ori, no stumbling blocks.)
(c) Il n’y a personne, ses volets sont baissés, pas de voiture (Thilliez
La Chambre des morts, 2005). (There’s no one there, his shutters
are down, no car.)
The above syntactic reduction creates an obvious effect: the expression

is anchored in the inner discourse of the character observing a situation
and evaluating if it presents danger.
(14) Motif [Q alors que P?] (6 occurrences)

(a) Que faisait-il au domaine accompagné de Victor, à boire des
canettes, alors que toutes les polices de France le recherchaient?
(What was he doing at the property with Victor, drinking from
cans, when all the police in France were looking for him?)
(Benacquista Tout à l’égo, 1999)
(b) Pouvait-on imaginer qu’il épargnerait des témoins, alors qu’il
avait étranglé une fillette innocente? Thilliez La Chambre des
morts, 2005). (Could we have imagined that he would spare wit-
nesses when he had strangled an innocent girl?)
This pattern expresses several values: the illogicality of a situation

(drinking from cans—Q—when on the run—P—where P, normally,
prohibits Q), or, on the contrary, its predictability (how could Q not
happen in the future when P has happened in the past and yet P is less
42
likely than Q?; P therefore “allows” Q). In a different form, because it

is more absurd (killing animals is less likely here than killing people),
Vargas exploits this motif whose form is no longer written Q whereas P,
but P then why Q?
(15) Je tue pas les bêtes, alors pourquoi je tuerais les gens? (I don’t kill
animals, so why would I kill people?). (Vargas Sous les vents de Neptune,
2004)
These few examples show that the construction grammar method allows
identifying patterns specific to a textual genre whose discursive function
or functions can be described. We believe that these patterns are con-
structions insofar as they correspond to a pairing between (variable but
identifiable) forms and meanings. Like the grammatical constructions,
the motifs can be schematic, semi-specific or even specific in the case of
phraseologisms.
6 Conclusion
This article maps out the notion of motif, albeit very incompletely
because many of its aspects are still blank or “uncharted.” Because of
the word’s polysemy and its varied conceptions, it is obviously impossi-
ble—and indeed undesirable—to argue that there exists an object des-
ignated by the word motif that all the disciplines reviewed here have
in common. Motif is a “crossroads” word, conveying a sense of which
disciplines can and should intersect to share concepts, methods and
tools. Beyond its primary purpose of presenting a heterogeneous notion
of a term (motif ), this article pleads for a collaboration between varied
perspectives. We are convinced it would facilitate refining certain defi-
nitions and identifying phenomena that certain approaches ignore. In
short, it would serve to account for shared elements, each with its own
specificities, that underpin the recurrent use of the word motif. Thus,
we started from narratology and folklore, which conceive of the motif
as a minimal unit. Initially independent of a particular form (the motif
is a kind of concrete theme), the motif acquired a “morphology” in the
43
Russian structuralist tradition, where it is linked to generic proposals

embodied in various statements but which are associated with defining
functions of the story. The “import” of motif into modern literary crit-
icism certainly is bearing fruit but can also be problematical: narrative
functions are in principle less stereotypical. However, at the level of the
analysis of literary genres as distinct as the crime or sentimental novel
we can detect characteristic patterns, be they recurrent lexico-grammatic
patterns (similar, in a way, to formal scientific patterns) associated with
a particular semantics, or abstractions giving rise to concrete achieve-
ments. Eventually, it will be appropriate for the linguistic analysis of
literary texts and for literary criticism itself to produce a higher-level
analysis, namely the study of the articulation of these minimum units
into more global units to assess the thematic developments of texts.
Substantial work beckons!
References
Aarne, Antti. 1961 (1910). The Types of the Folktale: A Classification and
Bibliography. Helsinki: The Finnish Academy of Science and Letters.
Abbott, H. Porter. 2008. The Cambridge Introduction to Narrative, 2nd ed.
Cambridge: Cambridge University Press.
Bertels, Ann, and Dirk Speelman. 2013. “‘Keywords Method’ Versus ‘Calcul
des Spécificités’: A Comparison of Tools and Methods.” International
Journal of Corpus Linguistics 18 (4): 536–60.
Boroda, Mojsej. 1973. “K voprosu o metroritmitsheski elementarnoj edinice v
muzyke.” Soboshtshenija AN GSSR 71 (3): 745–48.
Boroda, Mojsej. 1982 (1975). “Die melodische Elementareinheit.” In Sprache,
Text, Kunst: Quantitative Analysen, edited by Jurij Orlov, Mojsej Boroda and
Isabella Nadarejshvili, 205–21. Bochum: Brockmeyer.
Čech, Radek, Veronika Vincze, and Gabriel Altmann. 2017. “On Motifs and
Verb Valency.” In Motifs in Language and Text, edited by Haitao Liu and
Junying Liang, 13–36. Berlin and Boston: De Gruyter.
Christensen, D.L. 2003. The Unity of the Bible: Exploring the Beauty and
Structure of the Bible. Mahwah, NJ: Paulist Press.
Croft, William, and Alan Cruse. 2004. Cognitive Linguistics. Cambridge:
Cambridge University Press.
44
Darányi, Sándor, Peter Wittek, and Lázló Forró. 2012. “Toward Sequencing
‘Narrative DNA’: Tale Types, Motif Strings and Memetic Pathways.” In
Proceedings of Computational Models of Narrative 2012, edited by Mark Alan
Finlayson, 2–10. Cambridge, MA: Massachusetts Institute of Technology.
http://narrative.csail.mit.edu/cmn12/proceedings.pdf.
D’haeseleer, Patrik. 2006. “What Are DNA Sequence Motifs?” Nature
Biotechnology 24 (4): 423–25. http://146.6.100.192/users/CH391L/Handouts/
nbt0406-423-primer-whataremotifs.pdf.
Doležel, Lubomír. 1972. “From Motifemes to Motifs.” Poetics 4: 55–90.
Ducrot, Olivier, and Tzvetan Todorov. 1972. Dictionnaire encyclopédique des
sciences du langage. Paris: Éditions du Seuil.
Dundes, Alan. 1962. “From Etic to Emic Units in the Structural Study of
Folktales.” Journal of American Folklore 75 (296): 95–105.
Frontini, Francesca, Mohamed Amine Boukhaled, and Jean-Gabriel Ganascia.
2018. “Approaching French Theatrical Characters by Syntactical Analysis:
A Study with Motifs and Correspondence Analysis.” In Grammar of Genres
and Styles: From Discrete to Non-discrete Units, edited by Dominique
Legallois, Thierry Charnois, and Meri Larjavaara, 118–39. Berlin and
Boston: De Gruyter.
Ganascia, Jean-Gabriel. 2001. “Extraction automatique de motifs syntaxiques.” In
Actes de la 8ème Conférence sur le Traitement Automatique des Langues Naturelles
(TALN’2001). Tours (France), edited by Jean Véronis, Laurence Danlos, Pierre
Zweigenbaum, Nathalie Gasiglia, and Pascal Amsili (Bureau et Conseil d’Ad-
ministration de la ATAL), TALN’2001, Jean-Gabriel Gnascia, 1–10. Paris:
ATAL—Association pour le Traitement Automatique des Langues. http://tal-
narchives.atala.org/TALN/TALN-2001/taln-2001-long-017.pdf.
Goldberg, Adele. 2006. Constructions at Work. Oxford: Oxford University Press.
Gonon, Laetitia, Vannina Goossens, Olivier Kraif, Iva Novakova, and Julie
Sorba. 2018. “Motifs textuels spécifiques au genre policier et à la littéra-
ture « blanche ».” In Congrès Mondial de Linguistique Française, 9–13 juil-
let 2018, Mons, Institut de Linguistique Française, [onligne]: https://doi.
org/10.1051/shsconf/20184606007, Paris, 1–14.
Harris, Zellig, Michael Gottfried, Thomas Ryckman, Anne Daladier, and
Paul Mattick. 1989. The Form of Information in Science: Analysis of an
Immunology Sublanguage. Dordrecht: Kluwer Academic Publishers.
Köhler, Reinhard. 2006. “The Frequency Distribution of the Lengths of
Length Sequences.” In Favete Linguis: Studies in Honour of Viktor Krupa,
45
edited by Jozef Genzor and Martina Bucková, 145–52. Bratislava: Slovak

Academic Press.
Köhler, Reinhard. 2015. “Linguistic Motifs.” In Sequences in Language and
Text, edited by George K. Mikros and Ján Mačutek, 89–108. Berlin and
Boston: De Gruyter.
Kraif, Olivier. 2016. “Le Lexicoscope: un outil d’extraction des séquences
phraséologiques basé sur des corpus arborés.” Cahiers de Lexicologie 108:
91–106.
Legallois, Dominique, Thierry Charnois, and Meri Larjavaara. 2018. “The
Balance Between Quantitative and Qualitative Literary Stylistics: How
the Method of ‘Motifs’ Can Help.” In The Grammar of Genres and Styles:
From Discrete to Non-discrete Units, edited by Dominique Legallois, Thierry
Charnois, and Meri Larjavaara, 164–93. Berlin and Boston: De Gruyter.
Legallois, Dominique, Thierry Charnois, and Thierry Poibeau. 2016. “Repérer
les clichés dans les romans sentimentaux grâce à la méthode des ‘motifs’.”
Lidil. Revue de linguistique et de didactique des langues 53: 95–117.
Longrée, Dominique, and Sylvie Mellet. 2018. “Towards a Topological
Grammar of Genres and Styles: A Way to Combine Paradigmatic
Quantitative Analysis with a Syntagmatic Approach.” In The Grammar of
Genres and Styles: From Discrete to Non-discrete Units, edited by Dominique
Legallois, Thierry Charnois and Meri Larjavaara, 140–63. Berlin and
Boston: De Gruyter.
Mahlberg, Michaela. 2012. Corpus Stylistics and Dickens’s Fiction. London:
Routledge.
Muryn, Teresa, Malgorzata Niziolek, Alicja Hajok, Wojciech Prazuch, and
Katarzyna Gabrysiak. 2016. “Scène de crime dans le roman policier:
essai d’analyse lexico-syntaxique.” In 5e Congrès Mondial de Linguistique
Française (SHS Web of Conferences 27), edited by Franck Neveu, Gabriel
Bergounioux, Marie-Hélène Côté, Jean-Marie Fournier, Linda Hriba, and
Sophie Prévost, 06007: 1–14. Les Ulis Cedex A/London: EDP Sciences.
https://www.shs-conferences.org/articles/shsconf/pdf/2016/05/shsconf_
cmlf2016_06007.pdf.
Ofek, Nir, Sándor Darányi, and Lior Rokach. 2013. “Linking Motif Sequences
with Tale Types by Machine Learning.” In Workshop on Computational Models
of Narrative 2013, edited by Mark A. Finlayson, Bernhard Fisseni, Benedikt
Löwe, and Jan Christoph Meister, 166–82. Wadern: Dagstuhl Publishing.
http://drops.dagstuhl.de/opus/volltexte/2013/4150/pdf/p166-ofek.pdf.
46
Pike, Kenneth. 1967. Language in Relation to a Unified Theory of the Structure

of Human Behaviour, 2nd rev. ed. The Hague and Paris: Mouton.
Prince, Gerald. 2003. A Dictionary of Narratology, rev. ed. Lincoln: University
of Nebraska Press.
Propp, Vladimir. 1968 (1928). Morphology of the Folktale, 2nd ed. Austin:
University of Texas Press.
Quiniou, Solen, Peggy Cellier, Thierry Charnois, and Dominique Legallois.
2012. “What About Sequential Data Mining Techniques to Identify
Linguistic Patterns for Stylistics?” In Computational Linguistics and
Intelligent Text Processing. Part I, edited by Alexander Gelbukh, 166–77.
Berlin and Heidelberg: Springer.
Richard, Jean-Pierre. 1979. Microlectures. Paris: Seuil.
Sahraoui, Hedi-Théo, Pierre Holat, Peggy Cellier, Thierry Charnois, and
Sébastien Ferré. 2017. “Exploration of Textual Sequential Patterns.” In 14th
International Conference on Formal Concept Analysis: Rennes, France, June
13–16, 2017, Supplementary Proceedings, edited by Karell Bertet, Daniel
Borchmann, Peggy Cellier, and Sebastién Ferré, 99–102. Rennes: IRISA.
https://icfca2017.irisa.fr/files/2017/06/icfca_suppl_final_avec_isbn.pdf.
Siepmann, Dirk. 2005. Discourse Markers Across Languages. London: Routledge.
Thompson, Stith. 1946. The Folktale. New York: The Dryden Press.
Tomaševskij, J. Boris. 1925 (1965). “Thématique.” In Théorie de la littérature,
edited by Tzvetan Todorov, 263–307. Paris: Seuil.
Veselóvskij, Aleksandr. 1913a. “Poetika suzhetov i yeya zadatshi.” In Poetika.
Tom Vtaroj (Vol. 2). Poetika suzhetov (1897–1906), edited by The Imperial
Academy of Sciences, 1–8. St. Petersburg: Tipografija Imperatorskoj
Akademii Nauk.
Veselóvskij, Aleksandr. 1913b. “Motiv i suzhet.” In Poetika. Tom Vtaroj (Vol.
2). Poetika suzhetov (1897–1906), edited by The Imperial Academy of
Sciences, 9–12. St. Petersburg: Tipografija Imperatorskoj Akademii Nauk.
Vincensini, Jean-Jacques. 2000. Motifs et thèmes du récit médiéval. Paris:
Nathan.
3
Key Adverbs and Adverbial Motifs
in English Fiction and their French
Functional Equivalents
Iva Novakova, Dirk Siepmann and Marion Gymnich
1 Introduction
The present chapter starts from the observation that the automated
extraction of multi-word units, while being a worthwhile procedure for
identifying motifs, may blind us to other essential facets of literary lan-
guage or to the existence of synonymic motifs (“syngrams,” Siepmann
2015, 377, e.g. lost in reflection—absorbed in reverie—deep in thought)
whose lexical manifestations sometimes fail to exceed even a very liberal
I. Novakova (*)
D. Siepmann
M. Gymnich
University of Bonn, Bonn, Germany
e-mail: mgymnich@uni-bonn.de
https://doi.org/10.1007/978-3-030-23744-8_3
48
I. Novakova et al.
significance threshold. One such facet illuminated in pre-computer cor-

pus work (see Snell-Hornby 1983) is the dense use literary authors make
of a particular class of “descriptive” verbs,1 especially in English and
German; the situation in French is less well-researched (see Dyka et al.
2017). Such verbs encode actions and movement (or movement com-
bined with sound and light) as well as serving to characterize both main
characters and minor ones. Thus, for example, Gollum in Tolkien’s The
Lord of the Rings is characterized with descriptive verbs such as crawl, paw,
squeal, and hiss, all of which seek to arouse the reader’s suspicion and dis-
like, suggesting that he is more animal than human. The high keyness of
descriptive verbs is confirmed by modern corpus-based enquiry.
From verbs it is only a small step to adverbs, since the two word
classes commonly combine. However, although certain adverbs are
employed much more frequently in fiction than in other genres, they
have not yet been paid the attention they deserve. The present chapter
tries to remedy this situation by taking a detailed look at adverb use in
English and French fiction, thereby complementing the work done by
Snell-Hornby (1983) on verbs. Like Snell-Hornby’s investigation, this
study considers linguistic, literary, and translational aspects as well as
issues associated with the use of the word class under survey.
2 Method
The notion of keyword in its statistical sense (see Scott and Tribble
2006; see Stubbs 2010 for other uses of keyword) is by now well-
established in corpus linguistics. Words or word forms are said to be
“key” if they meet two requirements:
– A word, identified by the presence of a space or punctuation on

either side of it, is used with outstanding frequency in a particular
set of texts compared to a reference word list derived from a corpus
representative of the general language or a particular subset thereof.
Dunning’s Log Likelihood function (see Dunning 1993) determines
1Descriptive verbs appeal to the five senses: taste, touch, sight, smell, and sound.
3 Key Adverbs and Adverbial Motifs in English …
49
if a difference in frequency between two corpora is statistically

significant.
– The word must surpass a minimum frequency level. For the present
study, we used the following tool settings to determine it: minimum
frequency = 3, maximum key words = 2000; p value = 0.000001;
procedure = log likelihood.
Keyword investigations surface two kinds of evidence: (a) what the

text or texts being examined are about (“aboutness indicators”) and
(b) which word choices predominate in a stylistic sense (“stylistic indi-
cators”). The corpus-stylistic study of literary language is, by its very
nature, mainly interested in the latter.
This study is based on a keyword list of the English and French lit-
erary corpora compiled as part of the PhraseoRom project (all genres
combined; corpus LIT, see Appendix A); keywords were generated
by comparison with newspapers, scientific texts and TED talks that
we used as our reference corpus (corpus CONT, see Appendix A).
Additional examples were taken from a private collection of American
novels. The analysis proceeded in six steps. First, we manually identi-
fied all adverbs up to keyness rank 2000 ending in the suffixes *ly and
*ment as well as other adverbs. Next, we classified them according to
type (manner, time, place, frequency, etc.) before computing and com-
paring their relative and absolute frequencies shown in Table 1.
In the third step, we introduced a contrastive perspective while at the
same time narrowing the focus to manner adverbs, which intuitively
Table 1 Quantitative differences between different types of adverbs in English

and French
Types of adverbs EN FR
Overall 70 29
Manner 45 23
Time 5 4
Place 3 –
Frequency 1 –
Viewpoint 11 1
Degree 5 1
Total 140 58
50
I. Novakova et al.
appeared to pose the most serious translation problems and which, from
a stylistic point of view, can be expected to fulfil a primarily descrip-
tive function. Moreover, they are apt to contribute to an implicit char-
acterization of the novel’s personae. Thus, the choice of adverb may
have a considerable impact on the way a reader perceives a particular
character. The aim here was to explore qualitative differences between
potential translation equivalents, such as nervously—nerveusement or
slowly—lentement.
In steps four and five, we sought to identify what we termed “nat-
ural” French equivalents of some English adverbs. By “natural equiva-
lents” we mean French linguistic units that occur in a lexico-syntactic
environment closely resembling that which is observable in a similar
passage of an English-language novel. Space here precludes a detailed
discussion of the complex issues surrounding equivalence (for an
overview, see Kenny 2009). Suffice it to say here that equivalence
increases with the length of the items involved (see Siepmann 2014,
152). Bearing this in mind, we decided that the safest way of identi-
fying equivalents of adverbs was to start by classifying English manner
adverb + verb collocations by semantic category (e.g. cautiously + verb
of looking; briskly + verb of movement; step four). Step five entailed
exploring the lexico-syntactic environment of French verbs of look-
ing, motion, saying, etc. to identify adverbial expressions, which
involved a list of motion verbs from the DinaVmouv database (Stosic
and Aurnague 2017) as well as lists of other verb categories which we
generated by relying on dictionaries such as Le Petit Robert.
Finally, in step six, comparing natural interlingual correspondences
(or their absence) with actual translator behaviour is a natural follow-on
to the above contrastive analysis. Recently, a compelling case was made
by Kraif (2017, 58, following Rastier 2006) that “translationese” in no
sense differs from other forms of language variation since translated
texts become part of an intertextual body of language. We, on the other
hand, believe that it is precisely the comparison of original with trans-
lated texts that lets us ascertain to what extent the linguistic choices
made by translators become stock-in-trade for mother-tongue writers,
thus contributing to language change: “[la traduction] fait évoluer la
51
langue: le corpus des textes traduits s’intègre au corpus de la langue”

(Rastier 2006; cited in Kraif 2017, 58; [the translation] makes the lan-
guage evolve: the corpus of translated texts is integrated into the corpus of
the language; our translation). We hypothesize that this is probably
true at the macro-level of generic structures, but less so at the micro-
level of motifs. Moreover, we formulate two further hypotheses: First,
we suppose that in literary language, English employs more adverbs
than French, which often uses attributive adjectives instead of adverbs:
glance nervously = jeter un coup d’oeil nerveux. In Comparative Stylistics
of French and English: A Methodology for Translation, Jean-Paul Vinay
and Jean Darbelnet claim that English, due to its “synthetic character,”
favours adverbs and thus “allows a single word to be used where French
prefers a phrase, or even has no alternative. Not only do adverbs end-
ing in ‘-ment ’ seem cumbersome, they are restricted in their application.
Conversely, the suffix ‘-ly ’ in English can be attached to any adjective
and even to participles” (1995, 126). We will test this assumption by
applying statistical methods to large corpora. We further hypothe-
size that adverbs of manner (*ly or *ment ) tend to co-occur with cer-
tain semantic verb classes (speech, gesture, cognition, movement,
communication).
3 Results
3.1 Quantitative and Qualitative Differences
A fitting point for embarking on a comparison of key adverbs in

English and French resides in the fact that adverbs are generally more
frequent in English than in French—adverbs ending in *ment are only
about half as common as their direct English counterparts ending in
*ly. Unsurprisingly, this lopsided distribution is reflected in the propor-
tion of key adverbs found among the top 2000 keywords in English and
French in the aforementioned corpora. By far the largest class consists of
manner adverbs—39 in English and 23 in French. Interestingly, there
52
I. Novakova et al.
are fewer obvious correspondences in meaning between English and

French adverbs than one would expect (see Table 2).2
Other major numerical differences (see Table 1) concern viewpoint
or commenting adverbs (e.g. obviously 11 in English and franchement
1 in French) and degree adverbs (e.g. barely 5 in English vs. légèrement
1 in French). Since viewpoint adverbs tend to occur mainly in speech,
it might be hypothesized that French authors tend to represent spoken lan-
guage in accordance with somewhat different conventions than English
authors, but an exploration of this question is beyond the scope of this article.
A more detailed look at the relative frequencies of potential cor-
respondences (see Table 3) reveals equally striking discrepancies that
appear to provide some measure of support for the claim (to be investi-
gated below) made in contrastive stylistics that English adverbs are often
equivalent to “circumlocutory” structures in French (see Ballard 2003,
209; for a detailed pen-and-paper analysis of potential translation equiv-
alents in German–French translation, see Grünbeck 1976, 174–95).
More importantly, contrastive stylistics has overlooked the apparent idi-
osyncratic adverb use in each language. It may be hypothesized that nov-
elists are guided by language-specific predilections in this context on the
one hand but, on the other, may also have their personal preferences.
To compound matters, the items in Table 3 that appear to be full
equivalents usually differ in a number of uses or senses. In other words,
quantitative differences often shade off into qualitative differences:
slowly and lentement have similar uses with verbs of movement and ges-
ture as well as with verbs expressing light effects, but whereas English
slowly often occurs with speech verbs, collocations of the type “dire len-
tement” are comparatively rare in French novels (e.g. say + slowly: 10.3
per million; dire + lentement: 1.2 per million; the usual French equiva-
lent is “dire/prononcer d’une voix lente”).3 Silently is closely similar in
usage to silencieusement, except for its collocation with the speech verb
scream, which has no direct equivalent in French (but see constructions
2Columns 2 and 6 of Table 2 correspond to the decreasing keyness index obtained in
SketchEngine (see Kilgariff et al. 2014).

3For a discussion of verba dicendi and direct speech, see Chapter 4 by Diwersy et al. in this vol-
ume, which also stresses the productivity of the pattern say + Adv in the English literary corpus.
Table 2 English and French key manner adverbs with potential equivalents
1 33046.15 slowly lentement 27 2443.83 swiftly

2 14990.68 softly tendrement/ 28 2315.04 momentarily
mollement
3 14509.43 quietly 29 2294.18 brightly
4 11603.23 gently doucement 30 2292.10 wearily
5 8567.98 silently silencieusement 31 2291.19 dryly sèchement
3
6 7256.89 carefully 32 2210.31 absently distraitement

7 6434.74 abruptly 33 2204.96 sharply
8 4571.52 tightly 34 2166.08 neatly
9 4347.54 loudly 35 2133.60 violently
10 4224.86 thoughtfully 36 2096.07 briskly
11 3885.38 calmly calmement 37 2043.32 dimly
12 3861.16 angrily 38 2025.31 politely
13 3663.98 faintly 39 2024.31 helplessly
14 3633.03 vaguely 40 machinalement
15 3523.09 nervously nerveusement 41
16 3412.00 impatiently 42 rudement
17 3394.61 grimly 43 tristement
18 3061.81 cautiously 44 délicatement
19 3002.09 coldly froidement 45 gaiement
20 2965.83 perfectly 46 péniblement
21 2958.04 hastily précipitamment 47 gentiment
22 2794.84 lightly 48 sourdement
23 2741.90 intently 49 posément
24 2717.45 casually 50 négligemment
25 2532.41 wildly 51 furieusement
26 2525.07 stiffly
Key Adverbs and Adverbial Motifs in English …
53
54
I. Novakova et al.
Table 3 Quantitative differences between selected candidates for equivalence

(occurrences per one million words)
English key manner adverb French “dictionary equivalent”
slowly (200.41) lentement (23.68)
softly (71.40) tendrement (9.40)
doucement (88.34)
silently (41.82) silencieusement (1.56)
calmly (25.01) calmement (13.24)
nervously (20.89) nerveusement (2.13)
coldly (14.07) froidement (7.95)
absently (9.14) distraitement (0.82)
of the type “sa bouche s’ouvrit sur un cri muet”); calmly occurs far more
significantly in combination with wait than calmement does with atten-
dre. Absently frequently combines with speech and mental verbs, while
distraitement does not. Coldly is the only English manner adverb that is
frequently found alongside adjectives (coldly furious [see “rage froide ”],
logical, calculating, angry, formal, efficient, polite [see “politesse glacée ”]), a
use that has no direct parallel in French.
The English construction coldly + adjective may serve to illustrate the
possible implications of the differences between English and French use
of manner adverbs for their descriptive function in a literary text. The
lack of a direct parallel may have important repercussions on the way a
character’s demeanour or attitude is described, thus also posing a poten-
tial translation problem. In the collocation coldly polite, for instance, the
semantic modifier coldly can be expected to have a significant impact
on the reader’s impression of the character’s demeanour. While polite is
likely to give rise to a positive assessment of the character’s behaviour,
coldly polite is a much more expressive and intense way of describing
emotions and atmosphere (our “affective” discursive function [DFs]4,
see Appendix B); the construction emphasizes the distance between
characters interacting in a given scene and may convey attitudes rang-
ing from scorn to wounded pride on the part of the character whose
demeanour is described as coldly polite.
4In the PhraseoRom project, we distinguish between five basic DFs—that is, narrative, descriptive,
cognitive, affective, and pragmatic DFs—with some subcategories. For more details, see Appendix B.
55
The logic of frequency dictates that most of the differences noted so far
reveal infrequent uses or gaps in French rather than English. However, we
found a few instances that operated contrariwise: while both in English
and in French nervously is used with verbs expressing movement of the
hands, fingers,5 or feet as well as with verbs expressing smiling or laugh-
ter, verbs expressing a mental state appear to be more frequent in French
(e.g. craquer nerveusement [have a nervous breakdown], épuisé nerveusement
[nervously exhausted]); French froidement collocates with verbs expressing
the action of killing, where English resorts to set expressions (kill in cold
blood/cold-bloodedly); also, the use of froidement with accueillir finds no
direct parallel in English (give sb a frosty reception).
3.2 Adverb Use According to Verb Semantics

and Motifs Associated with Particular Uses
As should have become clear from this brief overview of qualitative dif-
ferences, some of the gaps found are lexicalized expressions (e.g. abattre
froidement—kill in cold blood ) which have been recorded in unabridged
bilingual dictionaries. Other gaps require more detailed investigation,
the major prerequisite being a classification of adverb use according to
verb semantics (step four in the methodology outlined above). Table 4
shows such a classification of manner adverbs.
Many of these combinations may be described as motifs, c ombining
as they do two lexical paradigms with a specific and recurrent
descriptive function (see Adam 2011, 267) in the fictional discourse.
As pointed out in the introduction, not all of the obviously literary
combinations listed in the Table 4 may, however, be identified by using
n-grams or lexico-syntactic trees. The combination of the motion verbs
stride, walk, march, and step with the adverb briskly appears to be char-
acteristically used to gender-stereotype the gait of a male agent as being
purposeful and swift. It suggests that he is (or is pretending to be) on
5See Chapter 5 by Grossmann et al. in this volume, which shows that V + nervously/nerveusement
is a very productive pattern in the context of the cigarette script.
Table 4 Distribution of manner adverbs by verb semantics (log dice >5)
Verb Gesture/ Motion Cognition Light/sound State/description Looking Speech
semantics facial
expression
56
Verbal nodes Frown, Rock, push, stride, Remember, Light, illumi- Arrange, fold, Observe, Say, reply,
with several smile, walk, march, nod, mock, wonder nate, glow, sit watch speak,
collocates laugh salute, tremble, flicker, gleam, continue,
shake, veer, turn, glimmer, announce,
stamp, trudge, burn, blaze, ask, remark,
shrug, climb, move, flash, glitter, comment,
I. Novakova et al.
stroll, gesture, cling sneeze, inhale add, mur-

mur, pause
Adverbial Lightly, Briskly, gently, Vaguely, faintly, Faintly, dimly, Stiffly, hastily, Cautiously Dryly,
collocates dryly violently, wearily, grimly, intently, brightly, casually, intently, thoughtfully
grimly, cautiously, violently, dimly, loudly tightly, neatly, sharply, (almost all)
swiftly, wildly, stiffly, carefully carefully, calmly
hastily, casually, calmly,
lightly, impatiently, violently
angrily, thoughtfully,
tightly, abruptly,
sharply
Specific Sniff, snore Gently curving; Glimpse
collocations loudly sharply dimly
defined;
tightly packed
wrapped/
clenched;
stiffly starched
57
an important and urgent mission (see also the following explanation in

the OED, s.v. “stride”: “often with implication of haste or impetuosity,
of exuberant vigour, or of haughtiness or arrogance”):
(1) Landow got out of his car and strode briskly into a large apartment
building. (Hammett The Assistant Murder, 1945)
(2) … the captain resumed his mask of boyish confidence. He walked
briskly to the radio control room. (Murphy Next of Kin, 2014)
(3) They both turned to glare as he strode briskly through the door. “Well,
gentlemen, sorry to keep you waiting,” said Phule. (Asprin A Phule and
His Money, 1999)
In these examples, the collocation stride briskly is syntagmatically

extended by means of a prepositional NP (into+NP). There is also a par-
adigmatic variation on NP (into a large apartment building, to the radio
control room, through the door ). These extended lexical units correspond
to our definition of motifs (see Chapter 1 by Novakova and Siepmann,
Sect. 3, in this volume) because they display lexical and syntactic reg-
ularities at the syntagmatic and paradigmatic levels and, at the same
time, are associated with particular narrative functions. Thus, the fact
that the construction consisting of a motion verb + briskly is combined
with adjuncts expressing a destination in the sentences above addition-
ally serves to stress the notion that the character acts purposefully. The
examples show that the widespread construction of literary characters
in accordance with gender stereotypes is often triggered by linguistic
items. While the decision to describe a character’s gait by employing the
manner adverb briskly may at first sight seem negligible for the over-
all understanding of a text, the frequent use of this adverb to refer to
the movement of male characters gives rise to a (comparatively subtle)
form of gendering that is anything but irrelevant since it reiterates the
traditional link between masculinity and being active. In other words,
the type of linguistic analysis suggested here proves to be a vital tool
for “approaches to fictional characters that include the category ‘gender’
in a systematic fashion” (Gymnich 2010, 506). To the reader, literary
characters typically seem to be more than a combination of units of
58
I. Novakova et al.
textual information, as Uri Margolin (1990, 463) points out: “Although

a possible individual is evoked or called into existence by a specific orig-
inating text, it is not reducible to words.” Yet, as the example discussed
here shows, lexico-syntactic features may give rise to a construction of
literary characters along the lines of gender stereotypes, for instance.
Linguistic input serves as a trigger for activating cognitive patterns that
are part of the reader’s general frame of reference, invoked for making
sense of the world.
Turning now to what we have called “natural” French equivalents
of motion verb + adjunct combinations, we find that French authors
opt for six regular choices, apart from the usually dispreferred adverbs
formed from adjectives by means of suffixation (see Table 5). While
the general syntactic choices in question have long been known to con-
trastive linguists and have been listed in some lesser known works (see
Grünbeck 1976 on German and French), the specific lexical paradigms
they constitute have never received detailed treatment. Table 5 gives a
detailed picture of these paradigms.
There are correspondences, within certain idiomatic constraints,
between columns 2 and 3, to the extent that the meaning expressed by
a noun following avec can sometimes be rendered by an adjective fol-
lowing a noun denoting a body part, a gesture or a movement, as in the
following clauses: il laisse retomber les bras d’un mouvement las (pattern
1, Table 5) = il laisse retomber les bras avec lassitude (pattern 2, Table 5)
(he lets his arms sink again with a tired movement).
Typically, however, each syntactic paradigm involves specific lexi-
co-semantic material, which in turn is dependent on the semantics of
the verb (see Table 4). French motion verbs display the widest range
of possible combinations, co-occurring with the six types of adjuncts
(Table 5). Somewhat surprisingly in view of what standard grammars of
French lead us to expect, adjuncts involving manière and façon are so
infrequent with motion verbs as to be negligible, which explains why
they are missing in the patterns described in Table 5.
In addition to these syntactic patterns, which admit wide variation
of lexis, a number of prepositional phrases exist that are more restricted
in terms of their collocability: à longs/grands pas, à pas feutrés, à longues/
larges/grandes enjambées/foulées, en quelques enjambées rapides, à amples
Table 5 Natural French equivalents of English adverbs
Verbs (1) V + d’un(e) (2) V + avec + NOM (3) V + ADJ in (4) V + avec + un(e) (5) en signe de * N (6) V + dans +
NOM + ADJ apposition + NOM + ADJ noun (+ADJ)
Motion verbs + (a) geste, (b) p
eine, effort, (c) furieux, (d) sourire, air, adieu, recon- mouvement,
(a) Gesture, ouvement,
m difficulté, lassitude, joyeux, regard; bruit, naissance, geste, bond,
movement, allure, bond, lourdeur, raideur; penaud, fracas; vivacité, acquiescement, salut
body part saut, coup; fougue, vivacité, heureux, hâte, vitesse, assentiment,
(b) Difficulty, main, pied, empressement, fourbu, rapidité, soup- approbation,
3
speed, bras, doigt; précipitation; bredouille, lesse, puissance; dénégation,

liveliness, lenteur, précaution, léger, ivre, obstination, joie, impuissance,
emotion méfiance, soin; souriant, regret, crainte, soumission, bien-
(c) Emotion, insistance, détermi- riche, humeur, perfec- venue, ignorance,
state nation; tendresse, radieux, tion, effort protestation,
(d) Gaze, facial curiosité, passion, malade, remerciement,
expression; fureur, inquiétude, insatisfait, apaisement,
sound, speed, mollesse hagard, reconnaissance,
liveliness, glorieux désapprobation,
emotion deuil, résignation,
encouragement,
négation
Looking oeil, air, regard
Speech voix, ton, rire force, gravité, douceur,
colère, feu, entrain
Cognition surprise, stupéfaction, satisfaction
soulagement, satis-
faction, certitude,
fierté
Sound/light blancheur,
rougeur, éclat,
lumière, lueur
Key Adverbs and Adverbial Motifs in English …
59
60
I. Novakova et al.
foulées (with long/grand strides, with muffled steps, with long/large,

grand strides, in a few quick steps, with wide strides); finally, we find set
expressions of the type à la hâte, à toute allure (courir, déambuler ) sans
rime ni raison or se lever comme un ressort (in haste, at full speed, (run,
wander) nonsensically, or get up like a spring).
3.3 Motifs of Motion and Light
This section is devoted to the search for equivalents for English and
French adverbial motifs. For reasons of space, we will limit ourselves to
two examples involving motion and light effects. In each case, we have
selected typical sample sentences illustrating a motif, for which we then
proceed to outline translation proposals, that is the sentences preceded
by (a), and authentic French examples (preceded by b, c, etc.).
3.3.1 Verbs of Motion + Briskly/Stiffly
Returning first to the aforementioned motif where a verb of motion

is combined with briskly, often with a view to gender-stereotyping
the behaviour of a male character, we are now in a position to suggest
“natural” equivalents (e.g. 1a, 2a, 3a have been translated by us and
1b, 2b, 3b are authentic examples from our French literary corpora).
Depending on whether the agent’s vigour, purposefulness (1a), haste
(2a), or arrogance (3a) are emphasized, we find equivalents such as the
following:
(1a) Landau descendit de voiture et entra d’un pas résolu dans un

immeuble.
(1b) Il avança d’un pas résolu vers les bouleaux à l’arrière de sa propriété
(Vann Désolations, 2016). (He advanced briskly towards the birchtrees at
the back of his property.)6
(2a) Il gagna la salle de contrôle radio en quelques enjambées rapides.
61
(2b) Il descendit de voiture, referma tout doucement la portière. Il se

glissa dans la trouée, gagna la terrasse en quelques enjambées rapides.
(Ndyaie Trois femmes puissantes, 2009) (He left the car, closed the door
very carefully. He crept into the hole, arrived at the terrace in a few brisk
strides.)
(3a) Ils se retournèrent tous les deux et le regardèrent fixement lorsqu’il
franchit la porte à toute allure. “Messieurs, désolé de vous avoir fait atten-
dre,” dit Phule.
(3b) Sans rien dire à aucun de ses serviteurs et de ses écuyers, sans même
prévenir Lubias, il franchit la porte à toute allure. (Blanchard Ami et Amile,
1985) (Without saying anything to one of his servants or his varlets, even
without warning Lubias, he walked through the door at full speed.)
It is evident from these examples that the translation of such motifs

from English into French requires great sensitivity both to the literary
context and the different shades of meaning conveyed by descriptive
expressions. This stands in sharp contrast to other types of motif, such
as transition markers, which have comparatively fixed equivalents (e.g.
his thoughts were interrupted as—il en était là de ses réflexions quand ). In
the opposite translation direction, on the other hand, what is needed
is an awareness of straightforward stereotypical combinations of, say,
motion verbs and adverbs (e.g. il avança d’un pas résolu vers les bou-
leaux—he walked/strode/headed briskly towards the birches ).
At this point, the search for equivalences could be broadened to
include other adverbs, but an exhaustive survey is beyond the compass
of this study. We will confine ourselves to another fairly complex exam-
ple: the use of stiffly with motion verbs and adjectives expressing pos-
ture. Among the most significant collocates of stiffly, we find the verbs
walk, bend, rise, pose, nod, stand, sit, climb, march, clamber, and stride as
well as the adjectives erect and upright. Stiffly indicates difficult, clumsy,
or even painful movement, which is sometimes associated with military
drill, great age, anger, or fatigue. Here are some examples:
(4) He suddenly looked tired again. He sat down stiffly, on the sofa …
(Barker Darkmans, 2007)
62
I. Novakova et al.
(5) The Brig rose stiffly. He seemed to have aged ten years during the
night. (Smith Eagle in the Sky, 2006)
(6) I did not sleep again, but waited for morning, when I rose stiffly and
went to a workman’s café … (Banville Shroud, 2003)
(7) … he rose stiffly to his feet and stood by the chair, his body aching
from having sat so long. (Campbell Scared Stiff: Tales of Sex and Death,
1987)
(8) She walked stiffly, as if hurt somewhere deep inside, and met no one’s
eyes. (Hambly Sold Down the River, 2001)
The most natural French equivalents that suggest themselves are combi-
nations of type (2) in Table 5, with the nouns (avec ) raideur, lourdeur,
lassitude, difficulté, effort, peine ([with] stiffness, heaviness, fatigue, dif-
ficulties, effort, toil) being obvious choices but lourdement (heavily) is
also used frequently as equivalent:
(4a) Il s’assit avec lassitude sur le sofa …

(4b) Reprise soudain par sa douleur, elle s’assied lourdement sur le lit et se
remet à pleurer. (Clavel La Guinguette, 1997) (Again seized by her pain,
she sat down stiffly on the bed and started to cry.)
(4c) Braguette déboutonnée, il s’assit lourdement sur une chaise … (Queffélec
Désirable, 2014) (Fly unbuttoned, he sat down stiffly on a chair …)
(5a) Le brigadier se leva avec gravité / difficulté.
(5b) Elle se leva avec gravité. (Rochefort Encore heureux qu’on va vers l’été,
1975) (She stood up stiffly.)
(5c) L’autre se leva d’un air offensé, prit son chapeau rond à la main …
(Camus La Peste, 1947) (The other one stood up with an offended look,
took his round hat in his hand…)
(6a) De nouveau, je n’avais pas dormi, je me levai avec peine / pénible-
ment / pesamment / non sans peine / avec effort / avec lourdeur / raide…
(6b) Nicolas se leva avec peine du sofa et baisa Stépan Pokrovsky sur le
front. (Troyat Tant que la Terre durera-1 [1947], 2007) (Nicolas stood up
stiffly from the sofa and kissed Stépan Pokrovsky on the forehead.)
63
(7a) Il se leva les pieds raides … le corps endolori d’avoir été assis si
longtemps.
(7b) Il se leva avec un peu de peine, défroissa sa veste, et derrière le vitrage
vert de ses lunettes ses yeux paraissaient fatigués. (Jenni L’Art français de la
guerre, 2011) (He stood up a little stiffly, smoothed his jacket, and behind
his green-tinted glasses his eyes seemed tired.)
(8a) Elle marchait avec raideur / avec peine / péniblement.
(8b) Il vit entrer une femme d’un certain âge, les cheveux blancs coupés
au carré, qui marchait avec raideur au bras d’une jeune fille. (Modiano
L’Horizon 2010) (He saw a woman of a certain age enter, with her white
hair cut into a rectangular shape, who walked stiffly on the arm of a girl.)
While these fill the bill quite neatly for examples (4)–(8), this is not the
case for nod stiffly (example 9), which does not seem to have a natural
counterpart in French.
(9) Halleck allowed himself the thinnest of smiles, then nodded stiffly to
them and stalked away. (Abercrombie The Blade Itself, 2006)
Wordings such as acquiescer avec raideur, hocher sèchement la tête, hocher

douloureusement la tête or hocher lourdement la tête (nod stiffly, dryly,
painfully, or heavily) constitute one-off occurrences in our French cor-
pus; equivalents that could be constructed on the basis of translation
proposals recorded in dictionaries, such as hocher la tête avec froideur or
hocher la tête à contre-cœur (nod coldly or nod reluctantly) do not occur,
and the comparatively common construction se contenter de hocher la
tête parallels merely nod or nod briefly rather than nod stiffly. Thus, we
are dealing here with a motif that is quite commonplace in English fic-
tion but hardly ever found in French novels. Such lexical or “colloca-
tional” gaps (see Siepmann 2003) can be accounted for by either of two
hypotheses: (A) writers of fiction use certain motifs for no other rea-
son than that they are available in their language, or (B) in some cases,
writers in one language perceive “slices” of reality that writers in another
language fail to see or omit because they deem them irrelevant. (The
absence of a straightforward French equivalent of nod stiffly may tempt
one to look for an explanation in the realm of stereotypical national
64
I. Novakova et al.
characteristics; in other words, here the proverbial “British stiff upper

lip” comes to mind.) In this sense, translations and “translationese” that
render the motifs in question more or less literally may help readers who
do not know a text’s source language gain access to new perceptions.
There is also some rather compelling evidence for Hypothesis A in the
converse direction: Having a natural predilection for the adverbial struc-
tures listed in Table 5, French writers also extend these structures crea-
tively in ways that English is unlikely to adopt. Thus, we find se lever avec
une lenteur majestueuse (rise with majestic slowness)/avec une grâce trou-
blante (with disturbing grace)/avec une solennité bouffonne (with mock
solemnity) or hocher la tête avec une brièveté calculée (nod with calculated
brevity). While such expressions are systemically possible in English and
were indeed used in the eighteenth and nineteenth centuries, in modern
usage they are probably felt to be rather noun-heavy and generally over-
blown and thus tend to be avoided. In simple terms, we might say that
French authors like to play with the subtle shades of meaning which can
be conveyed by such prepositional phrases with adverbial meaning.
3.3.2 Verbs Denoting Light + Faintly, Dimly, Brightly
This brings us to our second example, the use of faintly, dimly, and
brightly with verbs denoting light, such as light, shine, glow, burn, gleam,
illuminate, blaze, flash, flare, glitter, sparkle, glint, glimmer, and shimmer.
Space does not allow us to treat this topic exhaustively, so we will con-
fine ourselves to a number of typical examples.
(10) There was a low fire glowing faintly on the hearth. (Burnett The Secret
Garden, 1911)
(11) He watched until the forest began to glimmer and then to glow
faintly as the stars went out. (Campbell Midnight Sun, 1990)
(12) There were a couple of people milling around, white T-shirts faintly
glowing against the fade of the light. (Boyle Talk, Talk 2006)
(13) The Exeter Hotel is a six-story limestone building in the middle of a
block of discount shoe stores and dimly lit bars. (Auster Man in the Dark,
2009)
65
(14) The kitchen in the empty lodge was large, dimly lit. (Abbott Panic,
2005)
(15) The coal of the cigar glowed brightly for a moment. (Banks Canal
Dreams, 1989)
(16) It was late, almost eleven o clock, and the rows of bulbs on the
superstructure of Chelsea Bridge glowed brightly in the navy-blue night,
… like the lights on a circus’s big top. (Boyd Ordinary Thunderstorms,
2009)
(17) It seemed in places that there were no breaks between the lights, as
if the city were a carpet of pure light, a fragment of the sun. The clouds
above the city glowed brightly. (Card Songmaster, 2002)
In physics, glow denotes light emitted by a substance or object at a high

temperature, which conveys a sense of its metaphorical meaning, refer-
ring to a warm, intense, steady light that emits heat without flames.
When combined with faintly, it would appear to imply less intensity
and warmth but still a fairly steady emission of light and heat. The pro-
totypical image here would be that of a fire that has burned to embers
like the fire in sentence (10), and this is where the search for French
equivalents might start. A trawl through the French corpus reveals that
there is no equivalent capable of conveying the full range of nuances
contained in the English collocation. Feu can be combined with bas or
maigre, which renders the lack of intensity quite well, but it rarely collo-
cates with luire and rougeoyer (glow), and even less frequently with luire/
rougeoyer + faiblement (glow + feebly) (no occurrences in our corpus;
two occurrences in French originals in Google Books, see example 10b),
a collocation which in any case fails to do justice to the connotations of
warmth and steadiness radiated by the English collocation.
(10a) Il y avait un maigre feu (rarely: un feu bas ) qui luisait faiblement
dans l’âtre.
(10b) Raynor la souleva telle une plume et alla la jeter rudement à côté
du feu de bois qui luisait faiblement au centre de l’antre. (Lazareff Atlantis:
Le Secret de l’Orphèdre, 2018) (Raynor picked her up like a feather and
66
I. Novakova et al.
threw her roughly next to the wood fire which glowed faintly in the middle
of the cave.)
(10c) Simplement un maigre feu dans l’âtre géant. (Claudel Les Âmes
grises, 2003) (Merely a small fire in the gigantic fireplace.)
In the next example (11) an unsteady, dim light (glimmer ) transitions

into a steady light (glow faintly). Renditions using such direct but weak
equivalents as luire faiblement/vaguement (glow weakly/feebly) are ruled
out because they cannot be used to express the difference between inter-
mittent and steady light. The noun lueur, however, enters into colloca-
tion with both faible (faint light) and vacillant or intermittent (flickering
or intermittent light) and may be coordinated with a more precise
descriptive noun such as rougeoiement (red glow). This is probably the
closest match we can find for this complex combination of two sensory
perceptions. Here too it should be noted that the French equivalents are
less commonly used than their English counterparts; in a sense, we are
dealing with a motif that is exclusively English:
(11a) Il resta là, (sous les étoiles qui s’éteignaient,) figé dans sa contem-
plation, jusqu’à ce que la forêt se mît à projeter une faible lueur vacil-
lante, suivie d’un dernier rougeoiement (au fur et à mesure que les étoiles
s’éteignaient).
(11b) La lueur vacillante du photophore à gaz éclairait vivement le sol …
(Vian L’Automne à Pékin, 1947) (The flickering light of the gas lantern viv-
idly lit up the ground …)
(11c) … le soleil a sombré laissant dans le ciel, après un dernier rougeoie-
ment, pendant un temps très court, une lueur diffuse de couleur impré-
cise, entre le rose et le gris. (Mosset Naufrage d’un amour sous les tropiques,
1999) (… the sun has gone down, leaving behind in the sky, after a last
red glow, for a very short time, a diffuse light of an uncertain colour, in
between pink and grey.)
Potential equivalents of example (12) admit some variation. Again, the

most straightforward rendition of glow faintly is luire faiblement, which,
as stated above, has the disadvantage of not capturing the specificity of
67
English glow. A search for a more precise equivalent could be based on

the noun lumière, which is commonly modified by colour adjectives
suggesting warmth, such as orange/orangé/rouge/doré/jaunâtre (orange/
red/golden/yellowish), and collocates with such verbs as diffuser, pro-
jeter, or refléter (spread, project, or reflect). A third possibility, suggested
by exploring the lexical field surrounding colour adjectives, is to com-
bine the participial adjective doré with a complement introduced by par:
(12a) Il y avait quelques personnes qui s’agitaient en tous sens. Leurs tee-
shirts blancs luisaient faiblement dans le crépuscule avancé. // Leurs tee-
shirts diffusaient une faible lumière dorée / orange / orangée / rouge /
jaunâtre // (avec) leurs tee-shirts dorés par les dernières lueurs douces du
crépuscule / par la dernière lumière du soleil.
(12b) Il regarda le ciel où la lumière rouge déclinait. (Clavel Le Soleil des
morts, 1998) (He watched the sky where the red light faded.)
(12c) Debout sur le sable doré par les rayons encore brûlants du soleil,
Gilbert le regarde s’éloigner. (Clavel Pirates du Rhône, 1957) (Standing on
the sand that has been turned golden by the still glowing rays of the sun,
Gilbert sees him move away.)
Unlike the previous examples, dimly lit has two direct equivalents in
French (faiblement/pauvrement éclairé ):
(13a) L’Hôtel Exeter est un bâtiment en roche calcaire à six étages entouré
de solderies de chaussures et de bars faiblement/pauvrement éclairés.
(13b) Dans le compartiment pauvrement éclairé, Maigret l’avait mal vu.
(Simenon Maigret a peur, 1953) (In the dimly lit compartment, Maigret
had scarcely been able to see him.)
(14a) La cuisine qui se trouvait dans la maison vide du gardien était spa-
cieuse et faiblement éclairée.
(14b) Le hall était faiblement éclairé et le préposé dormait derrière
le standard. (Vian Les Morts ont tous la même peau, 1947) (The hall
was dimly lit, and the employee was sleeping behind the telephone
switchboard.)
68
I. Novakova et al.
The next three examples illustrate once more how contextual factors lead
to subtly differentiated equivalence relations between a standard literary
motif in English (glow brightly) and its counterparts in French. Sentence
(15) effectively captures the brief moment when a cigar lights up, look-
ing fleetingly like a miniature coal furnace. Although rare in French, this
motif could be rendered adequately by rougeoyer or by jeter une lueur vive.
(15a) L’espace d’un (bref ) instant, le tabac du cigare jeta une lueur vive. /
Le charbon du cigare rougeoya un instant.
(15b) La mèche rougeoya un instant. (Savatier L’Oeil du centre, 1962) (The
wick glowed brightly for a moment.)
(15c) Puis brusquement une bougie qui grésillait, jeta une lueur vive et
expira. (Benjamin La Table et le verre d’eau, 1947) (Then, suddenly, a can-
dle which hissed glowed brightly and expired.)
Whereas light bulbs are commonly depicted as glowing brightly in

English, French ampoule does not usually co-occur with adverb + verb
collocations; ampoule + éclairer/illuminer/briller (bulb + illuminate/
shine) in (16a) are used without adverbial modification. The colloca-
tional gap that arises as a consequence of this could be filled by resort-
ing to one of the more common combinations of the verbs in question
with adverbials, such as briller de mille feux (16b)/de tous ses feux/d’un
éclat d’or/d’un éclat éblouissant/… (to shine with a thousand lights (16b)/
with all its lights/with a golden glow/with a dazzling glow).
(16a) Il était tard, presque onze heures du soir, et les rangées d’ampoules
qui ornaient la superstructure du pont de Chelsea brillaient / s’illumi-
naient / rayonnaient de mille feux / de tous leurs feux / d’un éclat d’or
dans la nuit bleu marine, semblables aux lumières qui éclairent un chapi-
teau de cirque.
(16b) La Grande Ourse brille de mille feux. (Vialatte Dernières nouvelles de
l’homme, 1978) (The Big Dipper glows in a thousand lights.)
Sentence (17) basically illustrates the same contrast between English

and French. Whereas English writers regularly portray the interaction
69
between clouds and the sun with verbs such as glow or dim (the light),
French writers think twice about using such constructions. A faithful
rendition could be based on the verb flamboyer (“avoir l’éclat d’une
flamme”) or rutiler (“briller d’un rouge ardent, d’un vif éclat”) (17a);
an alternative rendition would shift the focus from the verb towards the
noun, which could be modified by a colour adjective or noun (orange,
rouge ) (17a, 17d). Taking this type of solution further, we also find
more daring equivalents such as nuage incendié (17b, 17c)/ en feu.7
(17a) On eût dit qu’à certains endroits il n’y avait aucune solution de
continuité entre les lumières, comme si la cité était un tapis de lumière
pure, voire un fragment du soleil. Les nuages au-dessus de la ville flam-
boyaient / rutilaient. // Au-dessus de la ville flottaient des nuages d’un
rouge vif / incendiés.
(17b) Des nuages incendiés m’entraînaient ailleurs. (Geramys Le Reste du
monde, 1987) (Glowing clouds took me with them to another place.)
(17c) Une monumentale presqu’île de nuages incendiés surgit à l’horizon
dont la splendeur fragile et fugace forçait la pensée vers d’autres voies.
(Duras Moderato cantabile, 1958) (A huge near-island of glowing clouds
suddenly appeared on the horizon, whose fragile and fleeting splendor
forced one’s thoughts in other directions.)
(17d) Enfin, sans qu’il fût question de lueur du couchant, en plein midi
arriva un nuage rouge, exactement rouge comme un coquelicot. (Giono Le
Hussard sur le toit, 1951) (Finally, without any hint of sunset glow, at
noon a red cloud came that was as red as a poppy.)
The insights gained in this section collectively suggest that motifs are
often language-specific, posing problems for translators which we
will turn to in the next section. The collocational gap which has been
detailed above proves to be very interesting for the analysis of texts
from a literary studies perspective. After all, references to light (or a
lack thereof ) are of the utmost importance for describing settings and,
7John D. Gallagher, personal communication.

70
I. Novakova et al.
specifically, for evoking a particular atmosphere. Descriptions of light

are also apt to elicit an emotional and aesthetic response from the
reader, for instance by referencing the sublime (La Grande Ourse brille
de mille feux. ). In genres like fantasy or science fiction, which gener-
ally rely very much on text-internal world building, descriptions of
light may also function as a potent world-building strategy. Cases in
point include classics such as Tolkiens’s The Lord of the Rings and Lewis’
Narnia series. Beyond that, as the two examples just mentioned also
illustrate, light, or more precisely the contrast between light and dark-
ness, as well as fire are among the most common symbols in literature.
Given the weight that references to light tend to have in literary texts,
we can also assume that authors do not choose expressions like the ones
listed above randomly and that translators do their utmost to fill collo-
cational gaps, even if this means that the passage will sound unfamiliar
to the readers’ ears.
3.4 Equivalents Found in Translation Corpora
In this section, we compare the equivalents identified during our

detailed comparison of original terms with those coined by translators.
3.4.1 Verbs of Motion + Briskly
There are 491 occurrences of verbs denoting motion or speech in com-

bination with the adverb briskly in our English–French translation cor-
pus (67,949,032 words, see Appendix A).
Verbs of motion (371 occurrences) constitute 75.5% of these com-
binations. The remaining 30% correspond to speech verb + briskly. The
translation solutions are consistent with those suggested above. Thus, in
combination with walk (57 occurrences), briskly is successfully rendered
by arpenter, marcher d’un bon pas, se diriger d’un pas alerte/à grands pas
(vers …), accélérer le pas, entrer d’un pas vif or partir d’un pas résolu (walk
with an alert step/with fast strides [towards…], accelerate the step, enter
with a brisk step or start with a determined step). The following exam-
ple illustrates a possible transposition:
71
(18) … and footsteps approaching briskly outside. (Estleman Kill Zone,

1991)
(18a) Des bruits de pas rapprochés résonnèrent derrière la porte.
The translation here is obviously target-oriented. The sentence under-

goes an important transformation: the verb (approached ) is translated by
an adjective (pas rapprochés ) in French and the adverb (briskly ) by the
French verb résonnèrent (resounded). Due to the less frequent use of
adverbs in French, these transformed sentence structures can be related
to the phenomenon of metataxis (see Tesnière 2015 [1959]). However,
since rapproché means “close in time or space,” the French sentence fails
to reproduce the gradual nature of the approach: des pas décidés/résolus
s’approchaient. The next example is somewhat exceptional in that French
here has a descriptive verb at its disposal that corresponds to an English
verb + adverb collocation. On the other hand, stride (briskly) and arpen-
ter are standard equivalents that the translator might have been aware of.
(19) … for the better part of half an hour we walked briskly along Connecticut
Avenue, neither of us speaking. (Cornwell All that Remains, 1992)
(19a) Nous sortîmes de son immeuble et arpentâmes Connecticut Avenue
pendant près d’une demi-heure sans prononcer un mot.
The solution proposed by the translator in (19a), again, is target-oriented.
3.4.2 Verb of Motion + Stiffly
Interestingly, the translation corpus comprises almost exactly the same

number of occurrences of the verb + briskly construction (490), distrib-
uted in about the same proportions as for briskly with verbs of motion
(83%) or speech verbs (17%). Walk stiffly (20 occurrences) is usually
translated quite felicitously by marcher (walk)/s’éloigner (move away)/se
diriger (go towards)/se lever (stand up)/traverser (cross)/monter (climb) +
avec raideur (stiffly)/d’un pas raide (with a stiff step). However, the fre-
quent translation of rise stiffly (14 occurrences) by se lever avec raideur
is not in keeping with the preferred way of putting this in authentic
72
I. Novakova et al.
French (se lever avec lourdeur / peine / effort; se lever péniblement get up
heavily/with toil/effort; get up painfully):
(20) Tabita rose stiffly from her chair. (Coe What a Carve Up!, 2008)
(20a) Tabita se leva de la chaise avec raideur.
Another interesting case discussed earlier (see Sect. 3.3.1) is the absence
of a French expression corresponding directly to nod stiffly (7 occur-
rences). Most translators fall into the trap of rendering this with a literal
hocher/incliner la tête avec raideur (nod/tilt the head stiffly) rather than
with attested equivalents such as hocher sèchement la tête, hocher dou-
loureusement la tête or hocher lourdement la tête, although acquiescer avec
raideur (nodding drily, nodding painfully or nodding heavily, nodding
stiffly) was found once in both original and translated works.
3.4.3 V of Light + Faintly, Dimly, Brightly
French translators consistently render glow faintly by means of the

verb + adverb collocation luire faiblement or by transposing this colloca-
tion to a noun + verb collocation (une faible lueur ). Examples:
(21) The monitor was glowing faintly around the edges. (Brown Digital
Fortress, 1998)
(21a) L’écran était noir …: une faible lueur était visible sur le pourtour.
(22) The surface of the sea glowed faintly with a reddish colour. (Jones
Divine Endurance, 1984)
(22a) La surface de la mer luisait faiblement avec une couleur rougeâtre.
(23) The clouds glowed faintly from within. (Pullman The Amber Spyglass,
2000)
(23a) Les nuages continuèrent à luire faiblement, de l’intérieur.
As we have seen, such renditions are perfectly acceptable: glowed faintly

(22) is translated by luisait faiblement avec une couleur rougêatre (22a).
Glow brightly, too, is rendered appropriately by rougeoyer when referring
73
to fire or embers in the translation corpus. The translator of the follow-

ing passage, however, appears to have been unaware of the connotations
of warmth that glow is imbued with:
(24) A street light glowed brightly, but without extending its light very far.
(Highsmith Ripley’s Game, 1974)
(24a) Un lampadaire projetait une lumière crue, mais dans un rayon
restreint.
Cru is glossed as “violent, brutal” in most dictionaries, so that projeter

une lumière vive (project a bright light) would have been a more appro-
priate choice. These examples provide further evidence suggesting that
in cases where English authors use adverbs of manner, their French
colleagues tend to prefer V + N + ADJ patterns such as projeter une
lumière crue (project a raw light). As most translators are aware of this
difference between English and French, the descriptive expressions
under discussion are generally rendered in a manner which conforms to
target-language norms (see also Dyka et al. 2017, 94).
3.5 Motifs Generated by V + Nervously
Let us look now at an English adverb that was found approximately ten
times more frequently in the English corpus (20.89 per million words)
than the corresponding adverb nerveusement in the French corpus (2.13
per million). This result confirms our hypothesis regarding the more
extensive use of adverbs in English (see also the previous sections).
In addition, the statistics obtained from our corpora for nervously/
nerveusement corroborate—in accordance with Hoey’s Lexical Priming
theory (2005) the preferential association between these adverbs and
certain semantic verb classes (e.g. verbs that refer to facial expressions
such as laugh, gaze ) or finger and hand movements (finger, fidget ). Thus,
nervously/nerveusement most often co-occurs with laugh (8.71 per mil-
lion), smile (7.51), giggle (7.80), rire (8.13), ricaner (7.80). The colloca-
tion laugh + nervously is used to introduce dialogues, thus assuming an
infranarrative DF (see Appendix B):
74
I. Novakova et al.
(25) Sheba laughed nervously. “He was supposed to phone an hour ago,
but I haven’t heard a peep yet. Wretched boy.” (Heller Notes on a Scandal,
2003)
(25a) Elle rit nerveusement.
– Et je me suis retrouvée avec deux haines en moi. (Rufin Katiba, 2010)
(She laughed nervously. —And I found myself with two objects of hatred
in myself.)
In both languages, the adverb nervously/nerveusement occurs preferen-

tially with verbs conveying finger movements: in English fidget (8.20),
finger (7.78), tap (7.70), caress (6.54), rub (6.43); in French tapoter
(9.05), pianoter (8.82), tripoter (8.54), feuilleter (8.19), frotter (8.38). In
English, fidget or finger + nervously appear most often in scenes where
security forces face a potential threat and are ready to shoot:
(26) He was fidgeting nervously with his pistols. (Forester Young

Hornblower, 1953)
(27) One of the soldiers nervously fingered the blunt-snouted tube of an
RPG-7, a Russian-made weapon almost certainly captured in years past
from the Iraqis. (Douglass Seal Team Seven-2, 2011)
Unlike in English, the combinations of verbs of finger movement (tapoter,

pianoter, feuilleter) and nerveusement in French typically reflect a character’s
emotional state and thus contribute to his/her implicit characterization:
(28) Brolin tapota nerveusement son paquet de cigarettes. (Chattam La

Trilogie du mal 3 Maléfices, 2004) (Brolin tapped nervously on his packet
of cigarettes.)
(29) Elle jeta sur la table un dossier sorti de sa serviette et le feuilleta
nerveusement. (Jonquier Du passé faisons table rase, 1982) (She threw a
folder from her briefcase on the table and leafed through it nervously.)
Furthermore, nerveusement co-occurs in French with verbs expressing

mental states (craquer/have a nervous breakdown 7.56, être épuisé/be
exhausted 7.29, sangloter/sob 7.37, étreindre/clasp 7.25). Such combina-
tions are rare in English:
75
(30) Parfois, lorsque ses journées de travail avaient été longues – et

elles devaient devenir, au fil des mois, de plus en plus longues – je la
sentais tendue, épuisée nerveusement. (Houellebecq Plateforme, 2001)
(Sometimes, when her working days had been long – and they had to
become, in the course of the months, longer and longer – I noticed her to
be tense, nervously exhausted.)
(31) Il a tenu trois jours puis a craqué nerveusement. (Lenteric La Gagne,
1980) (He persevered three days, then he broke down nervously.)
It is also interesting to note that in the parallel corpus, craquer nerveuse-

ment has been translated as having a nervous breakdown and not by
means of a construction using nervously, which proves that nervously in
English rarely or never co-occurs with verbs expressing mental states:
(32) She might have had a nervous breakdown. (Patterson Kiss the Girls, 1995)
(33) Peut-être qu’elle a craqué nerveusement.
However, in English, there are two uses of nervously that are conspic-
uously frequent: nervously + verbs of visual perception (glance [54
occurrences], peer, watch ); nervously + verbs of saying (e.g. say [27
occurrences], ask, mutter, murmur, suggest, whisper ). Such word combi-
nations are rare in French:
(34) The Grand Duchess glanced nervously at the officers standing in the
doorway to the cell. (Ash Prisoner of Ironsea Tower, 2004)
(35) “We thought you hadn’t come,” Mona said nervously but he didn’t
answer. (McGahern Amongst Women, 1990)
Unlike the laughter motif, which typically introduces dialogue (infrana-

rrative DF), the motif of looking appears often in descriptive passages
(often in coordinated [36] or juxtaposed [37] sentences where two or
more parallel actions follow one another):
(36) Ace hissed air out between his teeth and looked around nervously.
(King Needful Things, 1991)
76
I. Novakova et al.
(37) Nicole éteignait les cigarettes à peine entamées, tirait d’interminables

bouffées, parlait, se taisait, regardant nerveusement sur les côtés. (Queffélec
Les Noces barbares, 1985) (Nicole extinguished the cigarettes she had
hardly begun to smoke, took endless puffs, talked, fell silent, while look-
ing nervously at the sides.)
In this particular instance, the DF of the adverb nervously is both nar-

rative and descriptive. On the one hand, it conveys a sense of suspense;
on the other, it reveals the character’s emotional state (affective DF) and
enhances our understanding of his psychological make-up.
We turn now to how nervously/nerveusement is distributed in our
translation (parallel) English–French corpus. We find 834 occur-
rences (21.44 per million) of nervously and 585 (14.66 per million) of
nerveusement. The number of cases (70%) where nervously is translated
by its adverbial equivalent nerveusement is substantially higher than a
comparison of French monolingual corpora suggests, where nervously is
10 times more frequent than nerveusement. Here, the overall translation
behaviour is clearly source-oriented. Still, in 30% of the examples, the
translation is target-oriented, that is, nervously is rendered by different
periphrastic constructions. These equivalents are in accordance with the
systemic specificities of French. The most frequent French periphrastic
constructions follow the pattern V + N + ADJ.
When combined with nervously, verbs of visual perception tend to be
rendered by being transposed into a verb + noun + adjective combina-
tion of the type jeter un coup d’œil/take a look or un regard nerveux or
crispé/a nervous or tense look (see also Table 5):
(38) Kiukiu went in, glancing nervously around. (Ash Lord of Snow and
Shadows, 2003)
(38a) Entrez! Kiukiu obéit en jetant à la ronde des coups d’oeil nerveux.
(39) He nervously eyed the sea. (Ash Lord of Snow and Shadows, 2003)
(39a) Il surveilla la mer d’un oeil nerveux.
In combination with verbs of saying, nervously is most commonly trans-

lated into French by adjectival appositions (pattern 3, Table 5).
77
(40) “Take care, my lord,” Sosia said nervously. (Ash Lord of Snow and
Shadows, 2003)
(40a) Attention, seigneur! souffla Sosia, nerveuse.
(41) “Hi,” I said nervously. (Coe The Dwarves of Death, 1990)
(41a) Salut, dis-je, mal à l’aise.
(42) Bill said nervously. (King Insomnia, 1994)
(42a) … expliqua Bill, embarrassé.
All of these examples show the great variety of periphrastic expressions

used in French, most of which generate motifs. English prefers simpler
binary collocations (V + nervously).
4 The Restructuring of Sentences

Translators at times also restructure English sentences to bring them
closer to the periphrastic structures preferred in French. Thus, the
verbs laugh or glance, followed by nervously, are sometimes translated
by verbs like riposter or adresser (ripostai-je avec un rire nerveux [43a];
il m’adressait un regard crispé [44a]). In other words, the English verb
is transposed into a noun (un rire, un regard ) in French and the adverb
nervously into an adjective (nerveux, crispé ).
(43) I guess but I sort of showed you what it was about. I laughed nerv-
ously. – Oh, he said. (Meyer Revelation, 2008)
(43a) D’accord, mais je t’ai montré sur quoi il portait, ripostai-je avec un
rire nerveux. – Oh! Intéressant.
(44) As I got up I noticed him glance nervously at our table and light
another cigarette. (Barnes Talking it Over, 1991)
(44a) Je remarquai, en me levant, qu’il m’adressait un regard crispé et qu’il
allumait une nouvelle cigarette.
78
I. Novakova et al.
Simple English binary collocations (e.g. laugh or glance nervously) are

thus converted into a wide variety of patterns in French, as exemplified
by the sentences quoted above.
5 Differences in Frequency Revisited

To conclude, Table 6 summarizes the frequency of key adverbs in trans-
lation and comparable corpora.
The table indicates that the translation of adverbs from English into
French is a major source of “translationese.” Except for tendrement, we
find huge differences in frequency between authentic French fiction
and translated fiction, with adverbs such as silencieusement and lente-
ment being more than seven times as frequent in the translated texts.
This lends credence to our hypothesis (partly disproving Rastier’s (2006)
Table 6 Frequency of key adverbs in translation and comparable corpora

revisited
Translation corpora Comparable corpora
English key man- French “dictionary English key French “diction-
ner adverb equivalent” manner adverb ary equivalent”
slowly 8212 lentement 6689 slowly (200.41) lentement
(211.10 per (167.67 per million) (23.68)
million)
softly 2619 (67.32 tendrement 334 (8.37 softly (71.40) tendrement
per million) per million); douce- (9.40); douce-
ment: 4436 (111.19 ment (88.34)
per million)
silently 1524 (39.18 silencieusement 577 silently (41.82) silencieusement
per million) (14.46 per million) (1.56)
calmly 847 (21.77 calmement 965 (24.19 calmly (25.01) calmement
per million) per million) (13.24)
nervously 834 nerveusement 585 nervously nerveusement
(21.44 per (14.66 per million) (20.89) (2.13)
million)
coldly 568 (14.60 froidement 529 (13.26 coldly (14.07) froidement
per million) per million) (7.95)
absently 393 (10.10 distraitement 333 absently (9.14) distraitement
per million) (8.35 per million) (0.82)
79
assumption that translation “makes the language evolve”) that translated

texts do not have a major impact on the ways in which original texts are
written, at least at the micro-level of lexemes and constructions.
6 Conclusion
Drawing upon large corpora, our approach has allowed us to identify
both collocations and motifs that are specific to one of the two languages
under discussion. The vast repertoire of structures that we have described
could be used in lexicography, translation studies, contrastive stylistics,
and in creative writing. The large amount of data constitutes an input
for the lexicometric identification of motifs and, more generally, for dig-
ital stylistics. With respect to motifs, we observed significant syntagmatic
(a higher frequency of periphrastic adverbial constructions in French)
and paradigmatic variation, especially with respect to the different types
of verbs combined with adverbs in French and English. Our results
show that the motifs generated by manner adverbs are more complex
in French, which empirically proves higher creativity in French literary
language. The differences in creativity we observed could be explained
by systemic differences between the two languages. We were able to ver-
ify our hypotheses regarding the much more frequent use of adverbs in
English as well as preferences regarding combinations of adverbs and cer-
tain types of verbs (of motion, speaking, etc.). We hope to succeed in
verifying these results by using a French–English translation corpus.
References
Adam, Jean-Michel. 2011. Les textes: types et prototypes, 3rd ed. Paris: Armand
Colin.
Ballard, Michel. 2003. Versus: la version réfléchie. Paris: Ophrys.
Dunning, Ted. 1993. “Accurate Methods for the Statistics of Surprise and
Coincidence: Computational Linguistics.” Computational Linguistics 19 (1):
61–74.
80
I. Novakova et al.
Dyka, Susanne, Iva Novakova, and Dirk Siepmann. 2017. “A Web of

Analogies: Depictive and Reaction Object Constructions in Modern
English and French Fiction: Computational and Corpus-Based
Phraseology.” EUROPHRAS 10596: 87–101.
Grünbeck, Bernhard. 1976. Moderne deutsch-französische Stilistik auf der Basis
des Übersetzungsvergleichs. Teil I: Ordnungsliebe und logisierende Präzision als
Übersetzungsdominanten. Heidelberg: Carl Winter.
Gymnich, Marion. 2010. “The Gender(ing) of Fictional Characters.” In
Characters in Fictional Worlds: Understanding Imaginary Beings in Literature,
Film, and Other Media, edited by Jens Eder, Fotis Jannidis, and Ralf
Schneider, 506–24. Berlin and New York: De Gruyter.
Kenny, Dorothy. 2009. “Equivalence.” In Routledge Enyclopedia of Translation
Studies, edited by Mona Baker and Gabriela Saldanha, 96–99. London:
Routledge.
Kilgarriff, Adam, Vít Baisa, Jan Bušta, Miloš Jakubíček, Vojtěch Kovář, Jan
Michelfeit, Pavel Rychlý, and Vít Suchomel. 2014. “The Sketch Engine: Ten
Years On.” Lexicography 1: 7–36.
Kraif, Oliver. 2017. “Traduire le polar: une étude textométrique comparée
de la phraséologie du roman policier en français source et cible.” Synergies
Pologne 14. http://gerflint.fr/Base/Pologne14/pologne14.html.
Margolin, Uri. 1990. “The What, the When, and the How of Being a
Character in Literary Narratives.” Style 24 (3): 453–68.
Rastier, François. 2006. “La traduction: interprétation et genèse du sens.” In Le
sens en traduction, edited by Marianne Lederer and Fortunato Israël, 37–49.
Paris: Minard.
Scott, Mike, and Christopher Tribble. 2006. Textual Patterns: Key Words and
Corpus Analysis in Language Education. Amsterdam: Benjamins.
Siepmann, Dirk. 2003. “Collocations in Trilingual Perspective: New Evidence
from Large Corpora and Implications for Dictionary Making.” Cahiers de
Lexicologie 83 (2): 173–96.
Siepmann, Dirk. 2014. “Collocations Across Languages: Unity in Diversity?”
In Zweisprachige Lexikographie zwischen Translation und Didaktik, edited by
M. J. Dominguez, F. Mollica, and M. N. Curcio, 131–56. Berlin and New
York: de Gruyter.
81
Snell-Hornby, Mary. 1983. Verb-Descriptivity in German and English:

A Contrastive Study in Semantic Fields. Heidelberg: Winter.
Stosic, D., and M. Aurnague. 2017. DinaVmouv: Description, INventaire,
Analyse des Verbes de MOUVement: An Annotated Lexicon of Motion Verbs in
French. http://redac.univ-tlse2.fr/lexicons/dinaVmouv_fr.html.
Stubbs, Michael. 2010. “Three Concepts of Keywords.” In Keyness in Texts,
edited by Marina Bondi and Mike Scott, 21–42. Amsterdam: Benjamins.
Tesnière, Lucien. 2015 (1959). Elements of Structural Syntax. Amsterdam:
Benjamins.
Vinay, Jean-Paul, and Jean Darbelnet. 1995. Comparative Stylistics of French
and English: A Methodology for Translation. Amsterdam: Benjamins.
4
Speech Verbs in French
and English Novels
Sascha Diwersy, Laetitia Gonon, Vannina Goossens,
Marion Gymnich and Agnès Tutin
1 Introduction
Dialogue is one of the “typical narrative mode[s]” (Neumann and
Nünning 2008, 34) and a virtually indispensable structural feature of
S. Diwersy (*)
University Paul-Valéry Montpellier 3, Montpellier, France
e-mail: sascha.diwersy@univ-montp3.fr
L. Gonon · V. Goossens · A. Tutin
University Grenoble Alpes, Grenoble, France
e-mail: laetitia.gonon@univ-grenoble-alpes.fr
V. Goossens
e-mail: vannina.goossens@univ-grenoble-alpes.fr
A. Tutin
e-mail: agnes.tutin@univ-grenoble-alpes.fr
M. Gymnich
https://doi.org/10.1007/978-3-030-23744-8_4
84
S. Diwersy et al.
narrative texts. One factor which may help explain what makes direct
speech a standard feature of novels is that it enlivens narratives, a point
on which there is fairly widespread consensus among literary scholars. The
narratologist Monika Fludernik for one highlights this effect when she
argues that “speech representations are geared towards producing an effect
of vivacity” (Fludernik 2005, 559–60). In a similar vein, Rose-Marie
Weber opines that presenting “a character’s words in a narrative” serves
to make the text “lively, memorable, and often colourful” (Weber 2008,
558; see also Neumann and Nünning 2008, 105; Thomas 2007, 80). Not
only what the characters say produces this effect; it is also triggered by var-
ious linguistic features of direct speech, including exclamations, ellipses,
the use of colloquial expressions, regional or social varieties of a language,
verbal idiosyncrasies, the portrayal of speech defects, et cetera. Features of
direct speech like these may serve to individualize literary characters and/
or to depict them as representatives of a certain region or social class.
This implies that, beyond creating vivacity, direct speech typically
also contributes to the so-called reality effect, prompting the observa-
tion that “direct speech in the realist novel significantly enhances the
verisimilitude of the story” (Fludernik 2005, 562; see Page 1973, 3). In
non-realist genres such as fantasy and science fiction, direct speech may
render the world that is constructed within the narrative more complex
and plausible for the readers, for instance by endowing inhabitants of
different regions of an imaginary secondary world, such as Tolkien’s
Middle-earth, with distinct speech characteristics or “dialects”.
In narrative texts, direct speech fulfils a range of different func-
tions on the story level and in particular for figural characterization.
Dialogues have often been described as “advancing the plot” (Thomas
2005, 105; see Page 1973, 14; Neumann and Nünning 2008, 30).
Indeed, climactic moments and turning points in novels (for example,
quarrels between characters, declarations of love, confessions of guilt)
tend to be presented through dialogue. Yet, direct speech, contrariwise,
at times also slows down the action. This happens, for example, when
characters are made to engage in idle talk—possibly as part of a strategy
for presenting a speaker as a bore, a nuisance or a gossip. This device
illustrates what is regarded as the first and foremost narrative function
of direct speech, that is, enhancing both explicit and implicit figural
4 Speech Verbs in French and English Novels
85
characterization (see Thomas 2005, 105). In the words of Norman Page

(1973, 14): the “more customary role [of direct speech] is to contribute
to the presentation and development of character; and here its strength
lies in its being more direct and dramatic than authorial exposition”. To
put it another way, what characters say about themselves and/or about
others as well as how they say it typically tells the reader much about the
speakers and may engage the reader’s interest in the characters: “A vital
aspect of how we remember and grow close to fictional characters is the
way they speak: their accents and dialects place them geographically and
socially, while their verbal idiosyncrasies and catchphrases help to make
them memorable, even endearing” (Thomas 2007, 80).
Even though direct speech is often seen as the “paradigmatic form of
verbal mimesis” (Aczel 1998, 479), it is important to realize that dia-
logue in narrative texts is in fact far from being mimetic. The reader’s
impression that direct speech is mimetic is part of the narrative illusion,
the reality effect. After all, “[f ]ictional dialogue is an artificial version
of talk, partly shaped by a variety of aesthetic and thematic intentions
and conventions” (Toolan 1985, 193), even if it may imitate real-life
conversations to a certain extent (see Fludernik 2005, 559). A further
characteristic of direct speech in narrative texts is that it is by defini-
tion embedded in the narrator’s discourse. Even in so-called dialogue
novels by authors such as Ivy Compton Burnett and Manuel Puig (see
Neumann and Nünning 2008, 88), which “foreground speech and keep
narrative input to a minimum” (Thomas 2005, 105), direct speech is
still embedded in the narrator’s discourse, which is visible at least in
speech tags. These are necessary for identifying the speaker but may ful-
fil various other functions beyond this basic one. This is precisely what
we are interested in: the constructions that mark the narrative embed-
ding of direct speech, that is, the speech tags or inquit phrases—a fea-
ture that has often been neglected by literary studies.
Speech tags may differ considerably in terms of how “rich” they are
with respect to their semantics, ranging from the simple inquit phrase
he/she said to elaborate comments by the narrator. They often serve to
provide what Page calls “stage-directions as to facial expression, move-
ment, gesture, etc. – the expressive accompaniments of speech” (Page
1973, 26). Even though the examples Page chose are more complex (for
86
S. Diwersy et al.
example “here she began to sob ”, Page 1973, 26; original emphasis), the
verb itself to a certain extent may also function as a “stage-direction”.
Thomas (2005, 105), for instance, distinguishes between speech tags
“providing paralinguistic and prosodic information (e.g., she whispered,
he rattled on, moving closer to her )” and those that “provide an evaluation
of the dialogue and/or the characters (e.g., she confessed, he said, unsure
of himself )”. To achieve a more detailed and precise analysis of speech
tags we have drawn upon the typology of speech verbs or verba dicendi
developed by Harras et al. (2004).
As stated at the outset, in this paper we aim to analyze patterns of
direct speech that differentiate literary genres (specifically romance,
crime and fantasy novels, abbreviated ROM, CRIM, and FY) in French
and English. Direct speech is obviously omnipresent in these genres and
plays a vital part on the story level, as exemplified by passages such as
these:
(1) Au FBI on appelle ça «technique proactive», expliqua Brolin à ses col-

lègues. (CRIM, Chattam L’Âme du mal, 2002) (At the FBI, this is called
“proactive methods”, Brolin explained to his colleagues.1)
(2) “This is a Walther 7.65-mm. automatic and the magazine holds eight
rounds,” Nield informed him in a conversational tone. (CRIM, Forbes
The Power, 1994)
(3) Qu’est-ce que tu racontes? s’alarma-t-elle. (ROM, Musso Demain,
2013) (What are you telling me? she worried.)
(4) “I was kind of thinking just me and you, but sure, if you’re into that,”
he grinned, that sneaky wink reappearing. (ROM, Kelk I Heart New York,
2009)
(5) Tous les philtres de Khanaor ne le guériraient pas, soupira-t-elle en
secouant la tête. (FY, Berthelot Solstice de fer, 1983) (All of the potions of
Khanaor would not heal him, she sighed, shaking her head.)
(6) “We’ve got to find it in its lair,” said Lady Ramkin. (FY, Pratchett DW
08 – Guards! Guards! 1989)
1All translations in this article are ours, unless otherwise specified.

87
As illustrated above, direct speech patterns may vary considerably, from

a syntactic as well as lexical point of view. In (1), (2) and (6), the verbs
are classical verba dicendi, while in (3), (4) and (5) the verbs express a
psychological stance or attitude and thus cannot be considered stand-
ard speech verbs. In her corpus-stylistic study of Charles Dickens’s fic-
tion, Mahlberg (2013, 82) observes that “[t]he character information
conveyed by Speech clusters becomes most apparent when linked to
other textual cues”. In the most basic sense, the verbs used in (3), (4)
and (5) can be regarded as providing further “textual cues”, while those
in (1), (2) and (6) essentially just mark the act of speaking. It seems
likely that literary texts make more frequent use of the semantically
“richer” types of verbs than do descriptive, non-fictional genres such as
newspapers. The latter can be expected to privilege neutral speech verbs.
More specifically, we assume that a comparative analysis of direct speech
patterns can help us grasp some of the stylistic properties of literary sub-
genres such as romance, crime and fantasy novels. We further assume
that there are likely to be significant differences between the patterns
of direct speech occurring in French and English fiction due to general
linguistic tendencies and different stylistic preferences. Because of the
high frequency of adverbs ending in -ly in English (see Chapter 3 by
Novakova et al. in this volume), we suspect that the verbs in English
speech tags will often be modified by adverbs to convey additional
information. French constructions, where adverbs are less common, by
contrast, may vary more widely with respect to the verbs that are used
for expressing semantic nuances. Our study will focus only on a single
specific syntactic pattern, one in which direct speech occurs before the
speech tag, as in examples (1)–(6) above.
In the next section, we present our methodology, before moving on
to a discussion of our study’s most prominent results in the article’s
remaining sections. Our methodology is based on corpus linguistics and
drew on the verba dicendi typology established by Harras et al. (2004).
With its focus on the semantics of verbs occurring in speech tags, our
approach differs from earlier corpus-stylistic work on the presentation
of speech in literary texts (e.g. Semino and Short 2004). We selected
this approach because it held out a particular promise for bridging the
gap between linguistics and literary studies.
88
S. Diwersy et al.
2 Methodology
2.1 Corpora and Data Extraction
Our study is based on three sub-corpora from the PhraseoRom corpus

in French and in English described in Table 1.
We annotated the French and English corpora syntactically with the
XIP parser (Aït-Mokhtar et al. 2002) and then queried the annotated
samples with regular expressions to extract direct speech and speech tags
by means of CQP (Evert and Hardie 2011). A random sampling based
on keeping a fourth of the query matches was performed, with a preci-
sion of 83.4% for the French corpus and 68% for the English corpus.
The following sentence, for example, was extracted in error because the
quotation marks did not enclose direct speech:
(7) Mon coup de téléphone, émanant soi-disant de «Marceau», a été un

fameux pavé dans la mare. (POL, Malet Des kilomètres de linceuls, 1955)
(My telephone call, which supposedly came from “Marceau”, caused
quite a stir.)
The following parameters were extracted:
– direct speech
– speech verbs (lemma and form) and pronouns (in particular reflexive
pronouns)
– metadata of the extracted items.
As shown in Table 2, the elements were then checked and encoded

manually for a large number of examples.
Table 1 Corpora used for the study

Genre Number of Number Number Number of
tokens (French of texts of tokens texts (English)
texts) (French) (English texts)
Fantasy (FY) 13,323,976 104 17,814,599 142
Crime (CRIM) 17,859,351 194 13,561,947 115
Romance (ROM) 9,802,410 112 13,295,474 96
89
Table 2 Occurrences of direct speech

Subgenre Number of examples Number of examples
(French texts) (English texts)
Fantasy (FY) 3051 3012
Crime (CRIM) 2757 1778
Romance (ROM) 1716 2309
2.2 Linguistic Coding of Speech Tags
To reiterate, we wanted to compare semantic variation in speech tags

employed in English and French fiction in the three subgenres romance,
crime and fantasy. We assumed that significant differences in how
speech verbs were used would be observed. To describe this variation,
we carried out a manual semantic coding of the data.
Several classifications of speech verbs have been proposed in French
linguistics. In a large corpus and dictionary study, Lamiroy and
Charolles (2008) observed that most speech verbs in French are tran-
sitive verbs. Moreover, they were able to show that a significant num-
ber of intransitive verbs, especially those associated with the mode of
expression or with attitudes, can be used in a transitive manner and as
speech tags that enclose direct speech. Note, however, that the study by
Lamiroy and Charolles is based on a semasiological approach, which is
less suitable for our purpose of language comparison.
Because of its onomasiological basis, we, instead, considered the
typology of speech verbs proposed by Harras et al. (2004) as particularly
well-suited for our purposes. Originally developed with a large number
of speech verbs in German, it distinguishes between ten main classes of
verba dicendi, which it then divides again into subclasses. In the follow-
ing listing, we will only focus on the main classes, as follows2:
Gen: generic speech verb—for example dire, parler, say, speak
(8) Tu commences sérieusement à me casser les pieds, toi, fait la belle

Mélissa. (ROM, De Buron Dix jours de rêves, 1982) (You are really begin-
ning to get on my nerves, said the beautiful Melissa.)
2However, not all speech verbs may be used in speech tags.

90
S. Diwersy et al.
(9) “Go and look for her,” he said. (CRIM, Haynes Into the Darkest
Corner, 2011)
Repr: verbs that convey an assertive/representative act (as defined by

Searle): for example, affirmer (affirm), constater (state), admit, agree
(10) Il s’est porté volontaire pour entamer les travaux dans la crypte,
expliqua le chef milicien. (FY, Grimbert Les Armes des Garamont, 1998)
(He volunteered for beginning the construction work in the crypt,
explained the militiaman in charge.)
(11) “Yes, it would have been desirable,” Joshua agreed. (ROM, Balogh
Simply Love, 2006)
Dir: verbs which convey a directive act (as defined by Searle): for
example, demander (ask), prier (entreat), ask, demand
(12) Seigneur, mon Dieu, cher doux Jésus, sainte Marie pleine de grâces,
implore la brave vieillarde, … (CRIM, Dard Béru-Béru, 1970) (Lord, my God,
dear gentle Jesus, holy Mary full of grace, prays the honest old women, …)
(13) “Will you listen?” Strappi demanded. (FY, Pratchett DW 31 –
Monstrous Regiment, 2003)
Com: verbs which convey a promissive/commissive act (as defined by

Searle): for example, promettre (promise), menacer (threaten)
(14) C’est quand vous voudrez, avertit l’ingénieur du son dans son micro.
(ROM, Bourdin L’Homme de leur vie, 2000) (It’s whenever you like, the
sound engineer announced into his microphone.)
(15) “No names,” Newman promised. (CRIM, Forbes Deadlock, 1990)
Expr: verbs which convey an expressive act (as defined by Searle): for
example, gronder (thunder), critiquer (criticize), saluer (praise), sigh
(16) Vous voulez que je monte à pied? plaisanta Jeanne. (CRIM, Brussolo
L’Enfer, c’est à quel étage? 2003) (Do you want me to climb up on foot?
Jeanne joked.)
91
(17) “You’re an amazing woman, Daisy,” he sighed. (ROM, Kelly Always

and Forever, 2004)
Decl: verbs which convey a declarative act (as defined by Searle): for
example, baptiser (baptise), stipuler (stipulate)
(18) Tu remportes ce combat, Kira, proclama le Chevalier guérisseur pour

que tous l’entendent. (FY, Robillard Piège au royaume des ombres, 2003)
(You win this fight, Kira, announced the Knight and healer so that all
could hear it.)
We did not find any examples of this class in the English sub-corpus.
Struct: verbs which are used to structure the interaction: for example,
répondre (answer), ajouter (add), reply
(19) Un de mes hommes l’a confié à tes beaux-parents, indique Djamila. Ne

t’inquiète pas. Nous avons quelques questions à vous poser madame, enchaîne
Fabre … (CRIM, Giébel Les Morsures de l’ombre, 2007) (One of my men has
told your parents-in-law about it in confidence, Djamila mentions. Don’t
worry. We have to ask you some questions, Madame, Fabre continues …)
(20) “And the Falls coffin is different from the others,” Ellen Wylie added.
“I’m not saying otherwise,” Rebus interrupted. (CRIM, Rankin The Falls,
2001)
Mod: verbs which refer to the mode of expression: for example,

balbutier (stammer), murmurer (murmur)
(21) Je vous l’avais bien promis, qu’on se la jouerait à la Coppola! gloussa

amèrement Phan Hong. (CRIM, Jonquet Ils sont votre épouvante et vous
êtes leur crainte, 2006) (I had promised you that we would play it like
Coppola! Phan Hong chuckled bitterly.)
(22) “You poor, naive goddess,” the reedy voice wheezed into her ear. (FY,
Pollock The Glass Republic, 2013)
92
S. Diwersy et al.
Med: verbs which are related to the medium of expression: for example,
écrire (write), tweeter (tweet)
(23) Il n’est pas rentré! lui téléphona la jeune femme. (CRIM, Brussolo
Le Nuisible, 1982) (He hasn’t returned! the young woman told him on the
phone.)
(24) “The Man Who Can’t Commit will not want you in his own
domain,” Jude was reading out as Shaz fiddled with the Pride and
Prejudice video to try to find the bit where Colin Firth dives into the
lake. (ROM, Fielding Bridget Jones: The Edge of Reason, 1999)
Obviously, some classification problems may occur when a speech verb

is polysemous or when a verb can be assigned to different classes, as
is often the case with Mod speech verbs. Many of these also have an
expressive value, such as grommeler/grumble or gémir/moan. Relying on
reference dictionaries such as Le Petit Robert or Le Trésor de la Langue
Française, we decided to consider Mod verbs as expressive verbs when
a specific expressive/emotive function was associated with the mode of
expression. Grommeler, for example, is often associated with anger, but
can be used in many other contexts as well (when an utterance is hardly
intelligible for instance) and thus will not be considered as an expres-
sive verb but as a Mod verb. By contrast, a verb such as tonner/thunder
has been considered an expressive verb because, beyond the parameter
of manner (speaking very loudly), anger and violence are closely linked
to this particular verb.3
2.3 Lexicostatistical Set-Up
To examine the distribution of the verb classes listed above in the two
languages as well as in the different genres, we produced mainly three
data sets:
3The Le Petit Robert defines it as: “Exprimer violemment sa colère en parlant très fort” (express
one’s anger violently by speaking very loudly).

93
– VxL based on the cross tabulation of verb class and language,

– VxG_L based on the cross tabulation of verb class and genre, and
– VxG+L based on the cross tabulation of verb class and language
combined with genre. This data set appears in Table 3, whereas
the other two can be found in the appendix (see Tables 7, 8, and 9
respectively).
Table 3 Frequencies of verb classes cross-tabulated by genre combined with

language
Verb class Genre + Language Total
CRIM.ena CRIM.fr FY.en FY.fr ROM.en ROM.fr rows
Com 5 22 4 19 7 20 77
Decl 0 0 0 1 0 0 1
Dir 262 387 147 418 269 266 1749
Expr 63 312 89 521 185 174 1344
Gen 1053 762 2144 416 1301 388 6064
Med 0 4 1 2 3 3 13
Mod 94 353 328 524 229 243 1771
Repr 115 286 109 391 120 200 1221
Struct 179 571 189 697 189 383 2208
Total columns 1771 2697 3011 2989 2303 1677 14,448
aThe category labels were conceived in accordance with the pattern [genre].
[language]; thus CRIM.en stands for English (en) crime fiction (CRIM), FY.fr for
French (fr) fantasy fiction (FY), ROM.en for English (en) romance fiction (ROM),
et cetera
We processed these data sets using quite canonical methods in the tradi-
tion of French textometry (see Lebart et al. 1998), that is, by focussing
on contrastive word specificities4 and Correspondence Analysis (CA).
Some of the results, which we will discuss in the next section, had us
resort to lexical variation measures as well.
4The method of contrastive specificities, as introduced by Lafon (1980), has roughly the same
rationale as the keyword method applied in the British tradition of corpus linguistics (see,
amongst others, Rayson 2003), but, in contrast to the latter, its computation is based on hyperge-
ometric distribution.
94
S. Diwersy et al.
3 Results and Analysis

The purpose of using Correspondence Analysis (CA) was to provide a global
assessment of the distribution of the verb classes in relation to the configu-
ration of literary genre and language. In applying CA to the VxG+L data set
(see Table 3) we obtained the results shown in Fig. 1).5 They suggest that
Fig. 1 CA applied to the data set obtained by cross-tabulating verb class

and the combination of language with genre (The verb class is shown in red,
the combined category of genre and language in blue. Circle size represents
contribution)
5Computation and plotting of the CA were done with the R packages FactoMineR (Lê et al.
2008) and explor (Barnier 2017).

95
(i) the main opposition in our data set, as manifested on the horizontal
axis,6 holds between the two languages and not the genres (see {FY.
en, CRI.en, ROM.en} on the right-hand side versus {FY.fr, ROM.
fr, CRIM.fr} on the left-hand side of the plot); this opposition par-
allels the contrast between generic speech verbs for one and the
other verb classes for another ({Gen} on the right-hand side versus
{Expr, Com, Struct, Repr, Mod, Dir} on the left-hand side);
(ii) a secondary opposition, shown on the horizontal axis, pertains to
English and French fantasy fiction ({FY.en, FY.fr}) in the lower
half of the plot and English crime fiction ({CRI.en}) on the upper
right quadrant; this opposition coincides with the contrast between
expressive and manner verbs ({Mod, Expr}) in the lower left quad-
rant versus (mainly) directive and commissive ({Dir, Com}) verbs
on the left upper quadrant.
The computation of contrastive specificities provides a more detailed

insight into the tendencies revealed initially by the CA. Table 4 shows
the highly significant overuse of generic speech verbs in English com-
pared with other verb classes in both languages. On the other hand,
Table 4 Contrastive specificities of verb classes by language

9HUE&ODVV /DQJXDJH
(QJOLVK )UHQFK
&20
'(&/
',5
(;35
*(1
0('
02'
5(35
6758&7
6It should be noted that this axis alone represents 93.3% of the variance in the data.
96
S. Diwersy et al.
Tables 5 and 6 document the configurations of both English and

French fantasy fiction in comparison to the other two genres (CRIM
and ROM), which result from different tendencies to over- and
underuse certain verb classes.
Perhaps the most striking finding is that the generic verb to say plays
a very prominent role in speech tags in the English corpus. This verb
accounts for roughly 63% of verbs occurring in speech tags in the
English data set. The pervasiveness of this semantically unmarked or
“neutral” speech verb suggests that the lexical variation in the verb slot
Table 5 Contrastive specificities of verb classes by genre in the French data set
9HUE&ODVV *HQUH )UHQFK
&5,0 )< 520
&20
'(&/
',5
(;35
*(1
0('
02'
5(35
6758&7
Table 6 Contrastive specificities of verb classes by genre in the English data set
9HUE&ODVV *HQUH (QJOLVK
&5,0 )< 520
&20
',5
(;35
*(1
0('
02'
5(35
6758&7
97
of the speech tag construction is significantly lower in English than in

French novels (Fig. 2).
The vocabulary growth curves (VGC)7 we derived from LNRE esti-
mations (Baayen 2001) obtained with the zipfR package (Evert and
Baroni 2007) indicate that the speech tag construction in the English
corpus is less variable or productive than its French counterpart when
it comes to its verb slot. We address possible explanations for this phe-
nomenon and the special role of the generic speech verb to say in the
following sections.
Our findings also caused us to take a closer look at the verb class pro-
files that are characteristic of the three genres in English and French.
Upon closer inspection, we found that the internal make-up of these
Fig. 2 Vocabulary growth curves for the verb slot of the speech tag construc-
tion in the English and French samples
7The vocabulary growth curves were computed by interpolation based on random samples of
7000 tokens for both languages.
98
S. Diwersy et al.
profiles differs considerably between the two languages. Nevertheless,

they yielded the same configurations of genre oppositions: while fan-
tasy novels primarily contrast with crime and, to a lesser degree, with
romance novels in both languages, they do so due to highly differenti-
ated tendencies in the over- and underuse of the nine verb classes.
In the French corpus, fantasy novels, for instance, have in common
a positive association with expressive, manner and representative speech
verbs and also share a noticeable underuse of generic verbs. The exact
opposite holds true for crime fiction. Notably, the romance genre presents
a rather balanced profile except in the case of expressive verbs, which are
underrepresented—a feature that the genre shares with crime fiction.
In the English corpus, fantasy fiction differs from the other two genres
by virtue of its positive association with generic speech verbs and manner
verbs, while being negatively associated with directive, expressive, repre-
sentative and structural verbs. Apart from the shared negative association
with expressive verbs, this profile is, once again, opposed to that of crime
novels. The overuse of expressive verbs is characteristic of romance nov-
els in English, which also converge with crime fiction in the overuse of
directive verbs and the underuse of generic verbs (compared to the fantasy
genre). On the whole, the romance novels in the English corpus exhibit
a more differentiated profile of verb classes than does French romance
fiction. However, they are closer to crime novels than to fantasy fiction,
which holds true for both languages. Several aspects of these complex
configurations undoubtedly deserve further investigation, but here due to
space constraints, we focus on only two: the overuse of the generic verb to
say in the English corpus and the particularities of fantasy fiction.
4 Some Qualitative Remarks

4.1 The Verb to Say in Speech Tags in the English
Corpus
A huge number of different verbs can potentially be used in speech

tags, as our corpus analysis shows. Presumably, novelists would deem
it highly desirable for stylistic reasons to have a large variety of these
99
verba dicendi to choose from. By and large, this seems to be the case for
the French corpus. The English corpus, by contrast, suggests that some-
what different conventions rule in English novels. Our corpus-driven
approach revealed that to say, the most “neutral” or general verb possible
in speech tags, occurs far more frequently in them than any other verb.
This finding correlates with Page’s discovery (in “pre-corpus days”) while
analyzing the verbs in inquit constructions in a Charles Dickens novel
that the “opening chapter of David Copperfield has returned eight times,
asked and cried five times each, exclaimed, faltered and resumed twice
each, and repeated, replied, sobbed, mused and ejaculating once each, as
well as said thirty-seven times” (Page 1973, 26).
Additional findings further corroborate the observation that English
and American authors tend to use say far more frequently than other
verbs in speech tags. In his article “Analysing Fictional Dialogue”
(1985), Michael Toolan provides a sample analysis of a passage from the
short story “Cat in the Rain” by Ernest Hemingway, who is famous for
his extensive use of dialogue. He showed how “the seeming emotional
and temperamental gulf between the couple … is conveyed by and
reflected in their stilted uncooperative talk” (Toolan 1985, 202). What
is striking in the passage quoted by Toolan (1985, 203) is that the verb
say in the past tense is used six times while the only other verb appear-
ing in a speech tag in this passage is ask—a similarly “neutral” verb. In
analyzing the conversation patterns in the selected passage, Toolan does
not comment on how direct speech is inserted into the narrative, but
our statistical findings highlight the clear preference for the verb say.
What might otherwise appear to be accidental findings in two texts, our
corpus-linguistic approach shows to be a pervasive pattern. Hence, the
extremely high frequency of say is not merely a feature of particular lit-
erary texts or authors. Our English corpus, consisting of novels written
since the 1950s, reveals that the preference for say in speech tags can be
regarded as a general stylistic feature of inquit phrases in English fiction.
These striking findings raise the question of how to account for
them. The frequent use of say in the English corpus suggests to us that
the stylistic ideal of lexical variation is not the prime consideration for
English novelists, at least not with respect to speech tags. In his study
100
S. Diwersy et al.
French Style. L’accent français de la prose anglaise (2016), Gilles Philippe

describes the English literary style as follows:
… l’Angleterre était alors (et elle le reste aujourd’hui) tout simplement

moins sensible que la France à la reprise d’un même terme en contexte
étroit. … le dogme de la non-répétition d’un même mot ne s’est jamais
imposé en outre-Manche, du moins jamais avec la vigueur qui est la
sienne en France depuis le XIXe siècle (Philippe 2016, 69) (England
was (and still is today) simply less sensitive than France with respect to
the repetition of the same term in a narrow context. … the dogma of
avoiding repetition of the same word never took hold on the other side of
the Channel, at least never with the vigour it has had in France since the
nineteenth century.)
Besides tying the obvious preference for the generic speech verb say to
general stylistic features typical of English as opposed to French literary
style, it may also be an attempt by English authors to keep the narrator’s
intervention and “narratorial idiom” (Aczel 1998, 472) to a minimum.
This hypothesis is supported by the ingrained notion that the narrator
is supposed to fade into the background whenever dialogues are pre-
sented: “Due to its alleged directness, the representation of direct speech
is often considered as a form of ‘showing’ in which the narrator seems
to disappear (as in drama). Hence, as a narrative mode of representa-
tion, speech is closely related to dramatic performance” (Neumann
and Nünning 2008, 34). English novels apparently often seek to cre-
ate the illusion of the characters’ utterances being presented unfiltered
by the narrator—an effect that is enhanced by toning down the narra-
tive framing with non-descript speech tags. Why comparatively general
speech tags such as “he said/she said” might be deemed desirable in a
narrative text, Page explains by also saying that they supposedly ren-
der the narrator’s presence less “obtrusive” (Page 1973, 3) and provide
“a ‘neutral’ background against which the eccentricities of the dialogue
… can be more clearly perceived” (Page 1973, 14). Seen in this light,
fairly non-descript verba dicendi may after all seem stylistically desirable:
“In its purest form a passage may consist so largely of direct speech, so
little diluted with other elements, as to resemble an extract from a play”.
101
(Page 1973, 25; emphasis added) It has even been argued that “[e]arly
novelists in the English tradition honed their techniques in competition
with the stage, many even writing for the stage themselves” (Thomas
2007, 81). In the twentieth century, audio-visual media have presuma-
bly become the prime model for highlighting dialogue in the novel and
simultaneously toning down the narrator’s presence.
4.2 Expressive Verbs and the Specific Profile

of Fantasy in English and French
Having earlier pointed out the striking differences between fantasy nov-
els on the one hand and romance and crime fiction on the other, we
owe this phenomenon a closer look. The discrepancies between the gen-
res when it comes to preferences for certain verb classes are all the more
interesting in light of the fact that French and English fantasy novels
do not prefer the same types of verbs. Even granting that English novels
in general use a very high number of Gen verbs, it is still remarkable
that a contrastive approach reveals them as more typical of fantasy fic-
tion than of romance or crime novels. Moreover, Mod verbs are also
especially frequent in English fantasy. French, by contrast, clearly prefers
Expr verbs, followed by Mod and Repr verbs. We will focus on these
patterns, as we seek to describe what sets fantasy apart from the other
two genres.
The high frequency of Expr verbs in French fantasy suggests a predi-
lection for specifying the speaker’s attitude towards the interlocutor/s or
the situation at hand, as the following examples illustrate:
(25) Ne faites pas de promesses que vous n’êtes pas en mesure de tenir!
cracha Lucia, défigurée par le courroux. (FY, Robert Le Dire des Sylfes,
2003) (Don’t make promises that you can’t keep! spat Lucia, disfigured by
rage.)
(26) C’est souvent comme ça, hélas! déplora le chat en clignant ses yeux
mordorés. (FY, Boisset L’Antichambre de Mana, 2005) (It’s often like that,
alas! deplored the cat, blinking his golden brown eyes.)
102
S. Diwersy et al.
By and large, in French expressive speech verbs seem to be character-

istic of popular fiction in general rather than of fantasy in particular:
there are far fewer verbs of this class in the corpus of French general fic-
tion. The primary reason for the high frequency of Expr verbs in pop-
ular genres appears to be an attempt at (over)emphasizing the emotions
and attitudes of the characters: when emotions are expressed by direct
speech, they are often articulated again and thus reinforced by Expr
verbs. This strategy of “doubling” the information appears to be a char-
acteristic of popular fiction: the narrator “translates” or disambiguates
the character’s emotional stance, sometimes almost seeming to summa-
rize the direct speech, thus guiding the reader’s interpretation. The use
of Expr verbs consequently serves to simplify the reading process by
eliminating possible ambiguities: the reader is guided in his/her inter-
pretation of direct speech, since the narrator informs him/her about
the feelings that the character articulates. This lack of ambiguity may
even respond to explicit requests on the part of publishers, as Camille
Emmanuelle, an author of erotic romances for a major publishing
house, points out: “Les émotions des personnages doivent TOUJOURS
être exprimées. Le personnage ne peut jamais dire quelque chose sans
que l’on indique comment il le formule” (2017, 53) (“The emotions
of the characters must ALWAYS be expressed. The character is never
allowed to say anything unless the author indicates how he formulates
it”). The important thing here therefore is to use not simply il dit or elle
dit, but to specify how the character dit: “dis-je, bouleversée/répondit-il
en regardant au loin/dis-je en souriant et en baissant les yeux” (I say,
upset/he replied, gazing into the distance/I say, smiling and lowering
my eyes). This disambiguation does not necessarily involve the use of
expressive verbs, of course: French romance novels typically use the Gen
speech verb dire followed by a circumstantial complement to express the
speaker’s attitude and emotions (Gonon et al. 2019). This appears to be
the reason why the romance genre is associated with a comparatively
high frequency of generic speech verbs in both French and English nov-
els; in both languages, to say/dire is often used with circumstantial modi-
fications in romances, as the following sentences exemplify:
103
(27) – Avec un super dessert. Tu veux qu’on se prépare un bon gros des-
sert? – Bien sûr, dit-elle en sautant de joie. (ROM, Musso Et après…,
2004) (With a great dessert. Do you want us to make a good, big dessert?
– Of course, she said, jumping for joy.)
(28) “– I guess I’ll see you in a bit?” Rose said, trying to force a smile.
(ROM, McNamara From Notting Hill with Love… Actually, 2010)
In both examples, the direct speech itself presumably leaves some room
for interpretation of the speaker’s feelings, but the speech tag serves to
make these emotions abundantly clear.
The overuse of Expr verbs in French fantasy presumably stems from
the fact that fantasy tends to be a “hyperbolic” genre, seeking to estab-
lish analogies with an epic style of writing and stressing the extraordi-
nary nature of its characters: the inhabitants of fantastic worlds speak
loudly, shout, and howl (see Chapter 7 by Goossens et al. in this vol-
ume). The characters’ style of articulation also reflects and accompanies
a tendency towards displaying physical violence inherent in a genre that
typically features epic fights, in which, more often than not, the future
of the entire world is at stake. This epic or hyperbolic style is likely to
cause a considerable lexical diversity when it comes to characterizing
types of behaviour and forms of communication that are out of the
ordinary. The unbridled imagination (reflected in the invention of sec-
ondary worlds, weapons, maps, quests, peoples, etc.) may thus also cor-
respond to a more varied and extensive lexicon:
(29) La lumière de l’Immortel se propagea rapidement à l’épée et il cessa

brusquement son chant. Devant lui, Onyx, toujours entouré d’un bou-
clier d’énergie, s’était immobilisé. – Qui t’a enseigné à te cacher dans des
objets matériels ? vociféra Abnar. (FY, Robillard La Princesse rebelle, 2004)
(The light of the Immortal spread rapidly on the sword and he suddenly
stopped singing. In front of him, Onyx, still surrounded by an energy
shield, had come to a halt. – Who has taught you to hide inside material
objects? Abnar roared.)
(30) J’étais à peine au sol que Falci écrasait mon poignet gauche sous sa
botte pour neutraliser ma dague et appuyait sa lame sur ma gorge. “T’as
104
S. Diwersy et al.
raison, jubila-t-il. Le bonnet change rien à l’affaire.” (FY, Jaworski Gagner

la guerre, 2009) (I was hardly on the ground when Falci crushed my left
wrist under his boot in order to neutralise my dagger and pressed his
blade against my throat. “Y’are right, he rejoiced. The cap doesn’t change
anything.”)
In French fantasy, the other frequently used classes of speech verbs

also reinforce the importance of expressing the speaker’s attitude,
such as verbs denoting the manner of articulation (Mod verbs, which
refer to how one expresses oneself and thus often also allude to the
speaker’s attitude) and representative verbs denoting an assertive act
(Repr, which reflect the speaker’s confidence). These verb classes
taken together reflect an over-representation of characters’ emotions
and attitudes in French fantasy. This suggests that fantasy authors
establish their characters not (only) by what they say (i.e., in accord-
ance with a discursive ethos that tends to be preferred in general fic-
tion, where characters are usually presented to a considerable extent
via their direct speech); instead, characters are just as much con-
structed by how they say something (i.e., by spelling out their atti-
tudes), which amounts to a prior ethos (Maingueneau 2014). As a
genre, fantasy is very much action-driven (see Chapter 7 by Goossens
et al. in this volume), and everything results from this premise. Thus,
the profile of French fantasy with respect to the classes of speech
verbs seems to be informed by an “ambition” that is thwarted by the
genre’s limitations: on the one hand, the lexical diversity of speech
verbs suggests that authors of fantasy seek to “write well” by varying
the lexicon. On the other hand, they do not emulate general fiction
by representing literary characters primarily through their utter-
ances, opting for disambiguation and a hyperbolic depiction of verbal
behaviour instead.
In English fantasy, on the contrary, generic speech verbs prevail (see
also Chapter 7 by Goossens et al. in this volume):
(31) “Courage is found in unlikely places,” said Gildor. (FY, Tolkien The
Fellowship of the Ring, 1954)
105
This is presumably largely due to the general stylistic preferences of

English authors, which we discussed above in Sect. 4.1. In the English
fantasy examples in our corpus, there are, in fact, many speech tags
following the minimalist pattern say + N/pronoun. There are, how-
ever, also numerous examples of the Gen verb being modified by an
adverb or by a more complex construction. Just as in the French ROM
examples above, these constructions typically serve to convey informa-
tion about the speaker’s attitude (which could otherwise be articulated
by an Expr verb):
(32) “They’re getting hungry,” said Lupin coolly, shutting his briefcase
with a snap. (FY, Rowling Harry Potter and the Prisoner of Azkaban, 1999)
Beyond the frequency of the “neutral” verb say in the English corpus
in general, a further factor may help to explain the prevalence of Gen
verbs in the English fantasy sub-corpus: possible differences with respect
to the primary target groups of the fantasy novels in the English and
French sub-corpora.
A considerable number of novels in the English fantasy sub-corpus were
written for children. This bias in the corpus is not accidental but pays trib-
ute to the fact that the history of British fantasy fiction since the 1950s
has been shaped to a considerable extent by children’s fantasy, exemplified
by the works of influential authors such as C. S. Lewis, Roald Dahl, J.
K. Rowling and Philip Pullman. While children’s fantasy often features
a large number of “fiction words” (see Chapter 6 by Kraif and Gonon
in this volume) and thus offers some quite unusual linguistic material,
lexico-syntactic structures, on the whole, are likely to be comparatively
simple in children’s fantasy (and in literature targeted at children in gen-
eral). This may go a long way towards explaining the prevalence of generic
speech verbs in fantasy in the English sub-corpus. If the French fantasy
sub-corpus displays a higher frequency of Expr verbs than the English
sub-corpus, this may also be related to a grown-up or young adult target
readership for the French fantasy novels. After all, aiming at an older target
readership often correlates with featuring romance plots, which presuma-
bly encourages the use of expressive verbs in French fantasy.
106
S. Diwersy et al.
In addition to the linguistic preferences of English fiction in gen-

eral, the predominance of generic speech verbs in English fantasy in
particular may also be due to stylistic preferences of the subgenre and
even of individual authors. Several scholars have observed that English
fantasy tends to favour a comparatively simple language. Mandala
argues that this use of “plain language” helps support the illusion that
the imaginary worlds created in fantasy literature actually exist (see,
e.g., Mandala 2010, 95). In her study of Terry Pratchett’s fantasy
novels, Pillière claims that this particular writer plays with the jux-
taposition, on the one hand, of “an elevated or archaic style”, which
many readers might expect to find in literary texts that are trying to
evoke pseudo-medieval settings, and, on the other, of “a more col-
loquial register, thus producing a comic effect” (Pillière 2015, 115).
Interestingly, the example chosen by Pillière to illustrate her argument
contains direct speech and two speech tags that exemplify the stylistic
clash quite well: “As the cauldron bubbled an archaic voice shrieked:
‘When shall we three meet again?’ There was a pause. Finally, another
voice said, in far more ordinary tones: ‘Well, I can do next Tuesday’”
(Pratchett, quoted from Pillière 2015, 115). While other factors enter
into producing the noticeable stylistic shift in the passage quoted
above (e.g., the juxtaposition of a reference to the famous speeches
by the witches in Shakespeare’s Macbeth and a more mundane reply),
the contrast between the Mod verb shriek and the Gen verb say at
least contributes to the overall effect. The Pratchett example hints at
a further potential function of Gen verbs in speech tags, while simul-
taneously suggesting that Mod verbs may be employed to produce a
dramatic effect.
The frequency of Mod verbs in English fantasy may also partially
reflect its target readership. Verbs that refer to how direct speech is
uttered by a character presumably appeal to the imagination of children,
or are at least deemed particularly appropriate for this target group:
(33) “Harry, don’t complain!” yelled Ron excitedly. (Rowling Harry Potter
and the Goblet of Fire, 2000)
107
(34) “You madman!” shrieked Grandma Josephine, shaking Mr. Wonka

so fast his head became a blur. (Dahl Charlie and the Great Glass Elevator,
1972)
The fact that English fantasy displays a predilection for Mod verbs,
even if these are significantly less frequent than Gen verbs, may also
stem at least partly from the fact that this class of verbs lends itself
to stressing the diversity of creatures that can typically be found in
this genre: readers presumably expect the speech of orcs to sound
different from that of elves or pixies. The following examples illus-
trate this particular function of Mod verbs in English and French
fantasy:
(35) “What?” croaked the gnome. (Colfer The Arctic Incident, 2002)
(36) “– Kreacher will not insult Harry Potter in front of Dobby, no he
won’t, or Dobby will shut Kreacher’s mouth for him!” cried Dobby in a
high-pitched voice. (FY, Rowling Harry Potter and the Half-Blood Prince,
2005)
(37) “This is the only way,” whispered Gollum (FY, Tolkien The Two
Towers, 1954)
(38) – Moi qui te pensais perdue pour toujours, croassa l’affreux oiseau.
(FY, Robillard Irianeth, 2008) (Me, whom you thought lost forever,
croaked the terrible bird.)
In these cases, Mod verbs contribute to the text-internal world building

that is inherent in fantasy fiction.
Our strategy of focussing exclusively on the verbs in speech tags
has certain limitations. One is that these verbs are often modified by
adverbs or other constructions:
(39) “– I shall sleep downstairs, in the kitchen,” said Lady Ramkin cheer-
fully. (FY, Pratchett Guards! Guards! 1989)
(40) “Legally?” I asked, smiling back. – Of course, legally,” said Jackson,
feigning annoyance. (CRI, Francis Crossfire, 2010)
108
S. Diwersy et al.
(41) “Oh look, it’s Pippa and the rest of the girls,” says Fiona with surprise.
(ROM, Potter Don’t You Forget about Me, 2012)
Here, the constructions modifying the Gen speech verb fulfil a

semantic function that is very similar to that of Expr verbs in speech
tags in French novels. The main difference between English and
French speech tags thus may not be a semantic but rather a struc-
tural one. English speech tags are more likely to feature analytical
constructions (following the pattern V+modifier, e.g. said in surprise,
she said reproachfully ) in place of their synthetic French counterparts,
where the meaning of the verb corresponds to the meaning of the
English V+modifier construction (e.g. s’étonna-t-elle, reprocha-t-elle ).
Thus, the differences in how Gen and Expr verbs are distributed in
English and French novels may also result from linguistic differences
rather than from purely generic distinctions: if one takes into consid-
eration how say is modified, French and English fantasy may not be
so far apart after all.
In the English romance, however, Gen+modifier constructions
do not seem to be sufficient to indicate emotions and relation-
ships between the characters. This genre departs from the overall
tendency of the English corpus of privileging Gen verbs because it
over-represents the emotions of the characters (the Expr verbs of
the English romance can be translated into French by dire followed
by a circumstantial complement as mentioned above). Even if fan-
tasy in the English tradition seems to have been much more shaped
by Gen, it also favours Mod (though to a lesser extent). These
verbs may serve to refer to a speaker’s emotions at least indirectly.
Crime novels, by contrast, make very little use of lexical diversity
where verbs in speech tags are concerned; this probably tallies with
an emphasis on being dead serious, an attempt at mimicking a cin-
ematographic style and a tendency to privilege the depiction of
raw, blunt characters. The analysis of the contrasting specificities
shows that generic verbs are over-represented in French crime fic-
tion and directive verbs (ask, beg ) in the English crime novels. Both
types of speech verbs arguably serve the purposes described above
very well.
109
5 Conclusion and Perspectives

In our study, we explored to what extent lexico-grammatical patterns
differentiate literary genres within and across the two languages English
and French while developing a suitable lexicostatistical methodological
framework for investigating this question. The results obtained concern-
ing the speech tag construction in English and French fiction suggest
that narrative genres (here: crime, fantasy and romance) in both lan-
guages differ considerably in the distributional configurations of speech
verbs belonging to different semantic categories. Moreover, we found
that the corresponding genres in English and French do not share the
same semantic profiles when it comes to using speech verbs. On closer
inspection, this appears to be due to two major factors: (i) the perva-
siveness of the generic speech verb to say in English, and (ii) the special
position of fantasy in contrast to the remaining genres. The frequent use
of say in the English corpus seems to be related to two stylistic char-
acteristics of the English novel. First, the stylistic ideal of lexical varia-
tion, prevalent in the French tradition, is not the prime consideration
for English novelists. Second, the preference for generic verbs can also
be interpreted as an attempt at minimizing the narrator’s intervention
to help create the illusion of “unfiltered” utterances by the characters.
Another salient fact we highlighted is the specificity of fantasy as
opposed to the other genres studied both in French and in English.
This finding is all the more interesting because the two languages do
not prefer the same types of verbs. The French fantasy corpus is dom-
inated by expressive verbs. Together with Mod verbs, which character-
ize the genre as well (though to a lesser extent), Expr verbs contribute
to the overemphasis on the characters’ emotions and attitudes, which
is a general feature of popular fiction. This tendency is reinforced in
fantasy by the epic and hyperbolic nature of the genre. The English
corpus is dominated by generic speech verbs, even more so for fantasy
than for the other two genres. Beyond the overall stylistic preferences
of the English literary language, this is probably also due to the target
audience of many works that have shaped the English fantasy tradition
being children.
110
S. Diwersy et al.
We hope that the present study prepares the path for further research,
both within the PhraseoRom project and beyond. For one, we could
not analyze all of the extracted data as thoroughly as we would have
liked due to the constraints of the article format. For another, many
findings and hypotheses, such as the greater diversity of speech verbs in
French compared to English, should be pursued in future studies. We
would also like to expand the scope of our initial study in several direc-
tions. Firstly, we would like to study other fictional genres, in particu-
lar those represented in the PhraseoRom corpus (i.e., historical novels,
science fiction and general fiction). Secondly, we believe that compar-
ing our findings with non-fictional text types (newspapers, for example)
would be worthwhile. Thirdly, we are eager to refine both the analysis
of the syntactic structure of speech tags (by including circumstantial
and other extensions in the analysis) and the semantic analysis of speech
verbs. Finally, we would want to compare those initial results to other
forms of direct speech.
Appendix
See Tables 7, 8, and 9.
Table 7 Frequencies of verb classes cross-tabulated by language

Verb class Language Total rows
English French
Com 16 61 77
Decl 0 1 1
Dir 678 1071 1749
Expr 337 1007 1344
Gen 4498 1566 6064
Med 4 9 13
Mod 651 1120 1771
Repr 344 877 1221
Struct 557 1651 2208
Total columns 7085 7363 14,448
111
Table 8 Frequencies of verb classes cross-tabulated by genre for English

Verb class Genre (English) Total rows
CRIM FY ROM
Com 5 4 7 16
Dir 262 147 269 678
Expr 63 89 185 337
Gen 1053 2144 1301 4498
Med 0 1 3 4
Mod 94 328 229 651
Repr 115 109 120 344
Struct 179 189 189 557
Total columns 1771 3011 2303 7085
Table 9 Frequencies of verb classes cross-tabulated by genre combined for

French
Verb class Genre (French) Total Rows
CRIM FY ROM
Com 22 19 20 61
Decl 0 1 0 1
Dir 387 418 266 1071
Expr 312 521 174 1007
Gen 762 416 388 1566
Med 4 2 3 9
Mod 353 524 243 1120
Repr 286 391 200 877
Struct 571 697 383 1651
Total columns 2697 2989 1677 7363
References
Aczel, Richard. 1998. “Hearing Voices in Narrative Texts.” New Literary
History 29 (3): 467–500.
Aït-Mokhtar, Salah, Jean-Pierre Chanod, and Claude Roux. 2002. “Robustness
Beyond Shallowness: Incremental Deep Parsing.” Natural Language
Engineering 8 (2/3): 121–144.
Baayen, R. Harald. 2001. Word Frequency Distributions. Dordrecht: Kluwer
Academic.
Barnier, Julien. 2017. Explor: Interactive Interfaces for Results Exploration. R
package version 0.3.3. https://CRAN.R-project.org/package=explor.
112
S. Diwersy et al.
Emmanuelle, Camille. 2017. Lettre à celle qui lit mes romances érotiques, et qui
devrait arrêter tout de suite. Paris: Les Échappées.
Evert, Stefan, and Marco Baroni. 2007. “zipfR: Word Frequency Distributions
in R.” In Proceedings of the 45th Annual Meeting of the Association
for Computational Linguistics, Posters and Demonstrations Session,
Prague, Czech Republic, 29–32, 25–27 June. Prague: Association for
Computational Linguistics.
Evert, Stefan, and Andrew Hardie. 2011. “Twenty-First Century Corpus
Workbench: Updating a Query Architecture for the New Millennium.”
In Proceedings of the Corpus Linguistics 2011 Conference. University of
Birmingham, UK. https://www.birmingham.ac.uk/documents/college-
artslaw/corpus/conference-archives/2011/Paper-153.pdf.
Fludernik, Monika. 2005. “Speech Representation.” In Routledge Encyclopedia
of Narrative Theory, edited by David Herman, Manfred Jahn, and Marie-
Laure Ryan, 558–63. London and New York: Routledge.
Gonon, Laetitia, Vannina Goossens, and Iva Novakova. 2019. “Les
Phraséologismes spécifiques à deux sous-genres de la paralittérature: le
roman policier et le roman sentimental.” In La phraséologie française en
questions?, edited by Brigitte Buffard-Moret, Salah Mejri, and Luis Meneses-
Lerín. Paris: Hermann.
Harras, Gisela, Edeltraud Winkler, Sabine Erb, and Kristel Proost. 2004.
Handbuch deutscher Kommunikationsverben. Teil I: Wörterbuch. Berlin:
Walter de Gruyter.
Lafon, Pierre. 1980. “Sur la variabilité de la fréquence des formes dans un cor-
pus.” Mots 1: 127–65.
Lamiroy, Béatrice, and Michel Charolles. 2008. “Les verbes de parole et la
question de l’(in) transitivité.” Discours 2. http://journals.openedition.org/
discours/3232.
Lê, Sébastien, Julie Josse, and François Husson. 2008. “FactoMineR: An R
Package for Multivariate Analysis.” Journal of Statistical Software 25 (1):
1–18.
Lebart, Ludovic, André Salem, and Lisette Berry. 1998. Exploring Textual Data.
Dordrecht: Kluwer.
Mahlberg, Michaela. 2013. Corpus Stylistics and Dickens’s Fiction. New York
and London: Routledge.
Maingueneau, Dominique. 2014. “Retour critique sur l’éthos.” Langage et
société 149: 31–48.
Mandala, Susan. 2010. Language in Science Fiction and Fantasy: The Question of
Style. London and New York: Continuum.
113
Neumann, Birgit, and Ansgar Nünning. 2008. An Introduction to the Study of

Narrative Fiction. Stuttgart: Klett.
Page, Norman. 1973. Speech in the English Novel. London: Longman.
Philippe, Gilles. 2016. French Style: L’Accent français de la prose anglaise.
Brussels: Les Impressions Nouvelles.
Pillière, Linda. 2015. “Language Variety in Terry Pratchett’s Fantasy Fiction.”
In The Ethics and Poetics of Alterity: New Perspectives on Genre Literature,
edited by Maylis Rospide and Sandrine Sorlin, 110–25. Newcastle upon
Tyne: Cambridge Scholars.
Semino, Elena, and Mick Short. 2004. Corpus Stylistics: Speech, Writing and
Thought Presentation in a Corpus of English Writing. London and New York:
Routledge.
Thomas, Bronwen. 2005. “Dialogue in the Novel.” In Routledge Encyclopedia of
Narrative Theory, edited by David Herman, Manfred Jahn, and Marie-Laure
Ryan, 105–6. London and New York: Routledge.
Thomas, Bronwen. 2007. “Dialogue.” In The Cambridge Companion to
Narrative, edited by David Herman, 80–93. Cambridge: Cambridge
University Press.
Toolan, Michael. 1985. “Analysing Fictional Dialogue.” Language and
Communication 5 (3): 193–206.
Weber, Rose-Marie. 2008. “The Shape of Direct Quotation.” The Reading
Teacher 61 (7): 558–64.
5
Alcohol and Tobacco Consumption
in English and French Novels Since
the 1950s: A Corpus-Stylistic Analysis
Francis Grossmann, Marion Gymnich
and Dirk Siepmann
1 Introduction
The depiction of alcohol and tobacco consumption is a pervasive, but
as yet largely undocumented and unexplored, feature of twentieth- and
twenty-first-century fiction. Whereas news media and everyday conver-
sation tend to draw attention to alcohol and tobacco consumption pri-
marily as (addictive) habits, usually dwelling on their negative effects on
human health, we start from the assumption that novelists often portray
F. Grossmann (*)
e-mail: francis.grossmann@univ-grenoble-alpes.fr
M. Gymnich
D. Siepmann
https://doi.org/10.1007/978-3-030-23744-8_5
116
F. Grossmann et al.
consumptive habits for very different reasons. Firstly, references to alco-

hol and/or tobacco provide information about characters by depicting
habits and (possibly) addictions as a means of (implicit) characterization
of both main and marginal characters. Especially with the latter, which
are less fleshed out in the text, references to their alcohol and/or tobacco
consumption presumably often evoke stereotypes. Secondly, references
to alcohol and tobacco habits may provide the framework (and setting)
for certain plot strands or even for an entire narrative, as in the case
of a story revolving around an English pub or a Parisian café. Thirdly,
references to alcohol and tobacco use may evoke a specific socio-eco-
nomic or cultural background; that is, the kind of alcohol a character
drinks may for example function as a marker of social class, or a type of
tobacco may allude to a particular profession (see Montémont 2009).
Apart from exploring these macro-functions of literary depictions of
alcohol and tobacco consumption, we will consider the micro-functions
performed by recurring narrative patterns or motifs involving drink and
tobacco in their immediate contexts, discussing, for example, a proto-
typical cigarette script found in a large number of novels that emerged
from our corpus-based analysis.
2 State of the Art

2.1 The Cultural History of Alcohol and Tobacco
Consumption
Literary references to tobacco are bound to be informed by the cultural

history of tobacco consumption. What is deemed appropriate when
it comes to tobacco use by men and women or members of a particu-
lar social class has undergone significant changes in the last centuries.
Snuffing, that is, inhaling tobacco through one’s nose, for example, is
a practice that was very popular among the European nobility from
the eighteenth century onwards. It developed into an elaborate social
ritual that involved extracting the snuffbox from a handbag or pocket
with the left hand, tapping it a few times with the fingers of the right,
opening the box and placing some snuff on the back of the left hand,
5 Alcohol and Tobacco Consumption in English …
117
bringing it to the top of the nostrils and sniffing it as far back into the
nose as possible (see Burns 2006, 122). These were the “very motions
upon which aristocrats had long been relying to convey the scorn for
the masses” (Burns 2006, 121), so that snuffing became strongly asso-
ciated with contempt—an association that lends itself to being drawn
upon in literary texts. The practice of chewing tobacco, by contrast, was
common among poorer classes in nineteenth-century Europe and even
more widespread in the United States. Charles Dickens, while travel-
ling there in 1842, found himself thoroughly disgusted by the sight of
American fellow travellers who had “yellow streams from half-chewed
tobacco trickling down their chins” (Dickens, cited in Burns 2006, 116)
or who kept him awake at night with their incessant chawing. As late
as the 1950s, public places in the United States still provided spittoons,
but the custom now survives only in some rural backwaters and hence
has clear associations with country life and a certain backwardness. Still,
chewing was once also positively connoted as a substitute for chatter
(see Burns 2006, 139).
Smoking cigarettes, which is the most common way of consuming
tobacco today, has been a widespread habit in both Europe and America
ever since mass production started in 1884. Although cigarettes orig-
inated in working-class quarters, where the poor wrapped cigar butts
dropped by the rich into scraps of paper, the new tobacco industry soon
succeeded in marketing the new product as being supposedly milder and
more refined than cigars (see Burns 2006, 129)—and thus in tapping the
female market. The rise of the cigarette was favoured by clever marketing
strategies, such as the provision of free matchbooks with exterior striking
surfaces by cigarette companies, but also by social factors, with the hec-
tic pace of the new metropoles offering workers little other respite from
work (see Burns 2006, 139). For the “New Woman”, who defied conven-
tional gender roles, smoking cigarettes became one of the markers of the
gender equality she was claiming in a social and political climate that still
denied women the right to vote at the turn from the nineteenth to the
twentieth century. The cultural background we have just sketched already
suggests some of the (positive) characteristics that smoking cigarettes
tends to be associated with in many literary texts from the twentieth
century. A. D. Harvey (2014, 83) even considers cigarette smoking to
118
F. Grossmann et al.
have been “one of the defining elements of twentieth-century style ” (original

emphasis). Throughout literature from the beginning of the twentieth
century to the present, characters that smoke cigarettes are often depicted
as being modern, self-confident and independent.
The social history of alcohol in Western Europe is both longer and
better known than that of tobacco, beginning already in antiquity. Thus,
it should be sufficient for the present purposes to recall the ambivalence
that tends to be characteristic of references to the consumption of alco-
hol: on the one hand, wine, beer and other liquors have always been asso-
ciated with conviviality, feasts and pleasure; on the other hand, alcoholic
beverages have often been linked with excess, dangers and addiction. The
nineteenth century exacerbated this ambivalence and introduced clear
demarcations with respect to different alcoholic beverages: while good
wines (often consumed at the same time as cigars) were a sign of social
success within the bourgeois class, and oenology was linked to gastron-
omy (especially in France), the bourgeois elites simultaneously sought to
protect the lower classes from the dangers of alcoholism (a term that was
coined by Magnus Huss, a Swedish doctor, in 1849). Temperance leagues
developed to prevent the lower classes from becoming addicted to alco-
hol. In literary texts, different types of alcoholic beverages may give rise
to different associations. In Robert Louis Stevenson’s The Strange Case
of Dr Jekyll and Mr Hyde (1886), for instance, the narrator informs us
that the lawyer Mr. Utterson, who is the embodiment of self-control in
this Victorian novella, “drank gin when he was alone, to mortify a taste
for vintages” (Stevenson 1979, 29). This somewhat surprising habit sug-
gests that the consumption of alcohol was seen very ambivalently in the
nineteenth century. Nevertheless, drinking wine remained highly valued
in France, while the beer industry thrived in the Germanic countries,
England and the United States, and other, stronger beverages, such as gin
or whisky, were also very popular. In France, the Green Fairy (absinthe)
was banned just before World War I when people were encouraged to
drink less wine to make sure that the “Poilus” could have more in the
trenches. Since the nineteenth century, literature has time and again asso-
ciated creativity and drunkenness, while at times also showing the horrors
of alcoholism (e.g., in L’Assommoir by Émile Zola, published in 1877 or
in Anne Brontë’s The Tenant of Wildfell Hall, 1848).
119
Just before and after World War II, writers as far apart as William
Faulkner, Henry Miller, Antoine Blondin, Françoise Sagan and Charles
Bukowski continued to refer to alcohol in their works to evoke the diz-
ziness of the senses in collective feasts, but also despair and the danger
of decay. One may wonder, however, whether there has been a decline
in “literary alcoholism” since the 1970s; in recent decades, excesses
involving alcohol might perhaps have been increasingly replaced by ref-
erences to the abuse of other drugs (see Lacroix 2001).
2.2 Alcohol and Tobacco Consumption from the Point

of View of Literary Studies and the Purpose
of the Present Study
The aim of this chapter is to explore how English and French novelists
have used references to the consumption of tobacco and alcohol in nov-
els written since the 1950s for a considerable range of different literary
purposes. Even if the consumption of alcohol and tobacco has been a
pervasive feature in literary texts since the 1950s, as our corpus-driven
approach has shown, it has not been examined systematically so far—
neither from a literary perspective nor from a linguistic one. References
to alcohol and tobacco have been a common feature of literary texts for
centuries and may fulfil various functions. Novels that were written in
the last few decades could thus already draw upon a long literary tra-
dition of references to smoking and drinking alcoholic beverages, as
the examples mentioned above already suggest. Moreover, audio-visual
genres such as the film noir have established and reiterated stereotyp-
ical images such as the cigarette-smoking vamp, which have become
part of the repertoire of recurring associations with smoking and drink-
ing in Western cultures. The functions of references to smoking and/
or drinking may, in fact, be interdependent, which means that attempts
at establishing clear distinctions between them may in some cases seem
somewhat arbitrary. Still, for the sake of clarity, we will outline at least
the primary functions that have been attributed to descriptions of the
consumption of alcohol and tobacco in previous studies.
120
F. Grossmann et al.
2.2.1 The Diegetic Function
The first important function of references to smoking and drinking, as

already suggested above, is certainly their use as setting or background
of the plot (e.g., in scenes set in the café or bistrot as typical convivial
places in French novels, see Montémont 2009). Occasionally, drinking
may even become a major theme of a narrative text, as Ciobica et al.
point out with respect to James Joyce’s Ulysses: “It seems as though
half of the time, the characters are either at the pub or on their way to
meet someone there and even when the scene does not take place in
one of the local pubs, many of the characters presented are most likely
to be under the influence” (Ciobica et al. 2015, 147). As this example
already suggests, the act of smoking or drinking is often associated with
a time of transition, allowing the character to move from one place to
another—and the reader to move from one action to another in the
story. As these functions illustrate, references to smoking and drinking
may contribute to the organization of the narrative, even if they are, of
course, descriptive as well. As pointed out by Riffaterre (1986, 282),
“description generates the narrative”. Since they are vital for the con-
struction of the story (diegesis ), the two functions sketched above consti-
tute what can be called the diegetic function (Genette 1972).
2.2.2 Empathy or Critical Distance:

The Affective and Social Functions
By virtue of being part of social rituals, smoking and drinking typ-

ically serve to create a mood or an atmosphere in a literary text.
Conventionally, different types of alcoholic beverages tend to be asso-
ciated with distinct social or affective discursive functions (alluding to
loneliness vs. sociability, sexual frustration vs. erotic stimulation, etc.;
cf. Rozenberg 2015; for a list of discursive functions, see Appendix B).
Moreover, both tobacco and alcohol have often been drawn upon as
symbols in literary texts. Tobacco is, for instance, used as a metaphor
in Raymond Chandler’s hard-boiled crime novels The Big Sleep and The
Long Goodbye (Harp 1988, 95–104, referred to by Emanuel 2005, 87).
In The Big Sleep, smoke is imagined as a screen separating the characters.
121
Both smoking and drinking are very accessible ways of depicting

the relationship between literary characters. As Montémont (2009, 9)
rightly notes, “la consommation collective d’alcool et le choix des lieux
qui la permettent sont … une manière de construire le tissu relation-
nel dans lequel s’inscrivent les personnages” (the collective consumption
of alcohol and the choice of the places where drinking takes place are
… ways of building the social network to which the characters belong;
our translation). A description of a character’s emotions by means of
detailing his/her drinking and/or smoking habits may evoke the read-
er’s empathy with a certain character, but may just as well establish a
critical distance, especially in satirical novels like those by the American
author Philip Roth in which he “emphasizes the changing traditional
family structure and the relationships between man and woman” (Kalay
2016, 3248). References to the consumption of alcohol may reveal a lot
in this regard, fulfilling an indirectly descriptive discursive function (see
Appendix B).
References to alcohol and tobacco frequently function as markers
of emotional states (in particular negative ones such as nervousness or
frustration). This affective discursive function (see Appendix B) may
play a central role in providing information about literary characters.
According to Onwordi (2003), “cigarettes remind us of the characters’
fragile emotional states”. In other words, the ways cigarettes are held,
crushed, or thrown away are typically read as providing information
about a character’s current feelings (cf. Gonon et al. 2019). This type of
reading, of course, draws upon cultural conventions and clichés, which
have been reiterated in countless texts as well as in audio-visual media.
Drawing upon Eder et al. (2010, 35), the interpretive process at work in
such cases can be described as follows: “The processing of information
in the top-down mode involves the activation of a knowledge struc-
ture, such as a schema or a category, stored in long-term memory; this
structure is initially triggered by a piece of textual information and will
then guide the further processing as long as possible”. Several catego-
ries or clichés may be activated by references to drinking (alcohol) and/
or to smoking: Cigarettes and cigars have, for instance, conventionally
been used as erotic signals in fictional texts. In his study of “the sexual
semiotics of smoking in Victorian fiction”, Grylls (2006, 15) argues that
122
F. Grossmann et al.
the cigar “arriving early in the century came to be associated with male
sexuality and that its significance was subtly altered by the growing pop-
ularity, later in the century, of the more refined cigarette”. A case in point
is Thomas Hardy’s late Victorian classic Tess of the D’Urbervilles (1891),
where smoking can clearly be read as a symbol of Alec D’Urberville’s
erotic interest in the protagonist Tess. However, smoking and/or drink-
ing can also become an expression of sexual frustration in a literary text,
as Houellebecq’s novels exemplify. Especially in scenarios associated with
sexual tension and seduction one also encounters gender stereotypes such
as the femme fatale who is seductively smoking a cigarette.
Even more than smoking, alcohol, or rather the type of alcoholic bev-
erage that is consumed, tends to function as a powerful marker of social
class (see Montémont 2009, 4). If it occurs without any further specifi-
cation in French novels, the expression red wine is usually indicative of
a working-class milieu. In the two novels by Duras that were studied by
Montémont, by contrast, champagne competes with whisky (including
whisky and soda) and cognac as luxury drinks typical of the colonial
society where the stories are set.
In the following, we will examine whether semiotic functions like the
ones described above can be shown to have played a significant role in
English and French novels in the period from the 1950s to the present.
2.2.3 The Identification Function:

Characters and Literary Genres
Literary references to alcohol and tobacco are also highly effective

emblems and identity markers of characters (see Levet 2012) because
they trigger conceptual frames readers have acquired either in real-life
situations or via other fictional texts and audio-visual media. In her
discussion of the oeuvre of French noir writer Leo Mallet, Emanuel
(2005, 87–88) emphasizes this identification function, which may be
particularly apparent when brands are referred to in a text. Mentioning
cigarette or alcohol brands also contributes to the reality effect by sit-
uating a text (more or less precisely) both temporally and spatially/
culturally. Manchette was one of the first French writers who used this
technique: “Que ce soient les marques des voitures, celles des armes ou
123
celles des boissons, le fait de citer un nom existant, connu et pouvant

être vérifié procure au lecteur à la fois une sensation de réel ou de réal-
isme, mais aussi une certaine connivence avec les personnages des his-
toires” (Bentolila 2016, 61). (“Quoting existing, known and verifiable
names – of car brands, weapons or drinks – provides the reader with an
impression of both reality and realism, but also creates a certain com-
plicity with the characters of the stories”, our translation.)
Alcohol and tobacco, however, are not only used to identify literary
characters; they also serve as markers that help readers identify liter-
ary genres. This observation essentially holds true for realist as well as
non-realist genres (such as fantasy and science fiction): while brandy
and bourbon are common drinks in hard-boiled crime fiction, the
characters in J. K. Rowling’s Harry Potter novels drink Butterbeer, and
Longbottom Leaf pipe-weed is smoked in Tolkien’s The Lord of the
Rings. Linking patterns of alcohol/tobacco consumption and specific
genres seems possible at two different levels:
– at the level of action, scripts defining types of behaviour that are

typical of a genre can be identified. The notion of scripts, borrowed
from psychology (Schank and Abelson 1977), has been embraced by
cognitive narratology (see Herman 2006, 2009, 2013), but, despite
its heuristic potential, has often been deemed comparatively vague.
A corpus-linguistic approach based on the identification of recurring
linguistic patterns may help to address this problem by providing
both tools and evidence and thus may prove to be an important con-
tribution to cognitive narratology.
– at the level of characters/character constellations: for many genres, one
may identify either recurring stock characters or iconic characters
whose habits of alcohol and/or tobacco consumption are among their
distinctive features. Smoking and heavy-drinking private investi-
gators in hard-boiled crime fiction come to mind as recurring stock
characters, while Conan Doyle’s pipe-smoking super-sleuth Sherlock
Holmes and the prototypical wizard, Tolkien’s Gandalf, who is wont
to smoke his pipe before making important decisions, are iconic char-
acters that have come to stand for a genre (detective fiction, fantasy)
and have been emulated, alluded to or parodied by many writers.
124
F. Grossmann et al.
To sum up, we have identified a number of possible semiotic and soci-

ological interpretations of fictional depictions of alcohol and tobacco
consumption that ought to be verified empirically, that is, by means
of corpus stylistics. So far, there has been little research on the stylis-
tic, linguistic or discursive characteristics of literary representations of
alcohol and tobacco consumption. In the following, we intend to study
the three (interdependent) main functions sketched above: the diegetic
function, which is linked to the construction of the narrative; the affec-
tive and social function, which contributes to creating the “atmosphere”
of a particular scene and fosters the reader’s empathy or produces criti-
cal distance, while situating the characters within their specific cultural
and social environment; the identification function regarding characters
and genres, which draws upon signals or emblems that help the reader
categorize a character or assign a text to a particular genre. It is obvi-
ous that these three main functions are connected and may overlap;
after all, the same textual/linguistic clue can be interpreted as (a) an ele-
ment establishing the narrative background, (b) a signal of membership
in a certain social group, or (c) a typical feature of a particular literary
genre. The functions derive from three complementary perspectives, all
of which prove useful for the interpretation of recurring linguistic ele-
ments that have been extracted from the PhraseoRom corpus.
3 Method and Corpora

For the present study we drew on a subset of the PhraseoRom corpus,
namely the following five genres (in French and English): GEN (gen-
eral fiction), CRIM (crime fiction, detective novels, thrillers), ROM
(romance novels), SF (science fiction) and FY (fantasy). Each of these
genres is represented by 5 million words in the corpus. The French part
of the corpus includes 30,000,536 words (336 texts) and the English
part 30,212,347 words (241 texts). The two sub-corpora can thus be
considered as suitable for a comparison. To check the degree of specific-
ity of the Recurrent Lexico-syntactic Trees (RLTs) extracted from the lit-
erary corpus (LIT), we compared their frequency in a corpus of contrast
(CONT) consisting of non-fictional texts (see Appendix A).
125
The first step was the extraction of the RLTs related to tobacco or
alcohol consumption. (For more details on the methodology used for
the extraction of the RLTs, see Chapter 1 by Novakova and Siepmann
in this volume.) The second step was a comparison of the most pro-
ductive RLTs in both languages and an analysis of their textual func-
tions. Finally, we conducted a qualitative study on one RLT (for each
language), contrasting its use in the different genres.
4 Results
4.1 The Cigarette Script in English
and French Novels
4.1.1 The Cigarette Script in English
Almost all of the most productive RLTs in the English sub-corpus

are related to the cigarette script and only very few to the cigar script.
Table 1 provides a list of the RLTs we identified (with the minimum
threshold of 10,000 for log-likelihood, to ensure their statistical signif-
icance). The dispersion (DISP) guarantees that the same RLTs appear
in different novels (e.g., a dispersion of 80 indicates that the same RLT
appears in 80 novels), meaning that the frequency of an RLT does not
result from a particular text’s or author’s style.
The Main Functions of Light a Cigarette

In a first interesting finding of our corpus-driven approach we discov-
ered that references to characters lighting a cigarette are far more fre-
quent than references to the actual act of smoking. We may, of course,
assume that lighting a cigarette implies the (subsequent) act of smok-
ing. Still, as our examples show, lighting a cigarette fulfils a number of
specific functions within narrative texts. The following three functions
appear to be particularly prominent:
126
F. Grossmann et al.
Table 1 Most frequent RLTs related to the cigarette script in the English
sub-corpus
RLTs Freq. LIT Freq. CONT Disp. LIT Disp. CONT Log-
likelihood
Light a cigarette 271 2 80 2 675,0524308
Smoke a 94 0 42 0 10000
cigarette
Have a cigarette 39 8 28 2 62,651311
Take a cigarette 38 0 20 0 10000
Smoke a cigar 21 0 18 0 10000
Offer a cigarette 19 1 16 1 10000
Light a cigar 15 0 10 0 10000
Roll a cigarette 13 0 10 0 10000
Blow smoke 12 0 12 0 10000
a. Indicating a period of reflection on the part of a character
(1) I sat down cross-legged next to her and drew a burning stick from
the fire, using it to light a cigarette. I thought about what had happened.
(Smith Veteran, 2010)
As the example above illustrates, lighting a cigarette is frequently associ-

ated with an interim period, allowing the character to move on from one
emotional state to another through introspection or reflection on previ-
ous events (cognitive discursive function, see Appendix B). Moments like
these may set up decisions or even turning points in the novel. Thus, in
terms of the plot, lighting a cigarette sometimes also indicates a transition
leading up to a new event or an action on the part of the character.
b. Indicating a state of tension or weakness on the part of a character
Smoking a cigarette—or even lighting one—is conventionally viewed as

a reaction to feeling nervous. Thus, the cigarette script often correlates
with depictions of a character trying to cope with a difficult situation
and, consequently, alludes to the character’s emotions, which classes it as
an affective discursive function (see Appendix B). In these scenarios, the
act of smoking a cigarette is sometimes combined with the consumption
of alcohol or other drugs, as the following example illustrates:
127
(2) Patrick lit a Turkish cigarette and asked the stewardess for another glass
of brandy. He was beginning to feel a little jumpy without any smack.
The four Valiums he had stolen from Kay had helped him face breakfast,
but now he could feel the onset of withdrawal, like a litter of drowning
kittens in the sack of his stomach. (St Aubyn The Patrick Melrose Novels:
Never Mind, Bad News, Some Hope and Mother’s Milk, 2012)
c. Lighting a cigarette as part of a seduction script
(3) Vanessa walked to the armchair and sat down, crossed her long legs,
and lighted a cigarette. (Banks The Reserve, 2006)
The idea that lighting a cigarette is part of a seduction script has been
amply popularized across different media and, more often than not,
comes with gender stereotypes such as the femme fatale or vamp.
Although this use of lighting a cigarette can indeed be found in our cor-
pus, it is less frequent than one might perhaps expect. A variation on
the scenario of a character lighting a cigarette in a seductive fashion is
two characters sharing a cigarette, which likewise draws upon the con-
ventionalized sexual symbolism of smoking:
(4) Gregory lit two cigarettes and gave one to Charlotte. (Faulks Charlotte
Gray, 1998)
(5) Francis reached into his pocket and lit his wife a cigarette. (Moyes The
Last Letter from Your Lover, 2010)
The Character in Context: Smoke a Cigarette and Take a Cigarette

The most generic expression associated with the cigarette script—
smoke a cigarette—also has many different uses. It is often employed in
describing a situation and, more specifically, in introducing a character
at the beginning of a more extensive description of his or her demean-
our as illustrated here:
(6) He was smoking a cigarette, holding it in an affected way, between the

second and third fingers of his left hand. (Black The Silver Swan, 2007)
128
F. Grossmann et al.
In this example, the act of smoking itself is clearly less important for
implicitly characterizing the smoker than how he or she holds the cig-
arette (descriptive discursive function; cf. Gonon et al. 2019). The
example shows that the cigarette script may involve descriptions of a
character’s body language. The expression take a cigarette may have sim-
ilar functions, but is usually also part of a more elaborate sequence of
actions encompassing an interaction between two or more characters:
(7) He took a cigarette from the silver case Avery passed to him. (Ballard
The Wind from Nowhere, 1961)
Just a Pause for Thought or Talk: Have a Cigarette

and Offer a Cigarette
Among the other RLTs that we extracted, have a cigarette stands out as
an expression that has no exact equivalent in French (apart from the
verb fumer/smoke ). This expression can be found in the “GN want con-
struction”, which expresses both a lack of something and its palliation
(cf. Muller 2013) in situations where a character needs a break or seeks
to clear his or her head:
(8) She already had troubles back at the Yard with her superior because of
his having been removed from the case. Except for the ban on smoking
in all office buildings she would have had a cigarette. Probably two. Then
Gleitmann said something to her. (Kerr A Philosophical Investigation, 1992)
We also find the collocation have a cigarette in descriptions of interac-

tions between two characters, especially when one of them asks for the
other’s permission to smoke:
(9) Does it bother you if I have a cigarette? (Mayle A Good Year, 2004)
(10) Do you mind if I have a cigarette? (Faulks Charlotte Gray, 1998)
The cigarette script occurs quite often at the beginning of an interac-

tion—a narrative function that is most easily seen in the use of the ver-
bal collocation offer a cigarette. Moreover, it is interesting to note that
129
it is often “another cigarette” that a character offers, asks for, or wants

to smoke: the English corpus contains 98 occurrences of the colloca-
tion another cigarette and 16 of last cigarette. Smoking several cigarettes
in a row or continuing to smoke may, on the one hand, reflect the
importance of sociability in a text; on the other hand, it may indicate
addictive habits and/or an ongoing state of nervousness and tension
(fulfilling a descriptive narrative function). The pattern of repeated or
continual smoking appears to be much more frequent in the English
corpus. In the French corpus, we found only 23 occurrences of autre
cigarette and 14 of dernière cigarette. This statement must, however,
be qualified, since we also found 24 occurrences of nouvelle cigarette,
whose equivalent (“new cigarette”) does not appear in the English cor-
pus, even though the collocation is possible. This may indicate a slight
difference between the two languages with respect to the repetition
motif.
What Kinds of Cigarettes or Other Types of Tobacco Are Smoked?

The cigarette motif encompasses a considerable amount of paradig-
matic variation with regard to what is smoked (see Gonon et al. 2019).
One possibility in this context, which may allude to the geographical
or cultural setting, is the use of brand names. There are comparatively
few cigarette brand names in the English corpus: Camel 19; Dunhill 7;
Rothmans 7; Gauloises 6; Gitanes 4; Craven 2; Sportsman 5; Marlboro
“Red” 1; Lucky Strike 1; Chesterfield 1. Alternatively, authors some-
times use adjectives to specify the type of cigar or cigarette smoked by a
character: Havana (10); Turkish cigarette (8); French cigarette (4). The
information which brand or type of cigarette/cigar characters smoke
may of course contribute to their implicit characterization (descriptive
discursive function). It triggers the readers’ real-world knowledge which
is likely to have been informed in turn by literary conventions, audio-
visual media and advertisements. The latter have arguably played a
major role in endowing both cigarettes in general and individual brands
in particular with stereotypical associations.
Although they occur rarely in the extracted RLTs, there are many ref-
erences to cigars in the English corpus (459 occurrences for cigar, and
130
F. Grossmann et al.
69 for cheroot ). With 287 occurrences, the term tobacco is also quite
frequent. It appears in various contexts, for instance in references to the
smell of tobacco (which can sometimes work as a clue in crime fiction),
or when the gesture of someone about to roll a cigarette, that is, the
act of taking tobacco from a box or shaking it from a pouch, is men-
tioned. Moreover, the collocation roll a cigarette is also frequent in the
corpus. Instead of using the terms cigarette or cigar, characters or nar-
rators sometimes use more colloquial expressions: fag (92 occurrences
after sorting), ciggie (18 occurrences), or cig (16 occurrences).
Despite its prominent role in non-fictional contexts in recent dec-
ades, the negative impact tobacco has on people’s health is rarely men-
tioned in novels. In the following example, the allusion to smoking as a
health risk fulfils an ironic function:
(11) Dead time until he arrived. Time to light a lethal, cancer-causing cig-
arette, time to fly in the face of the Surgeon General’s advice – as if you
could trust a man who was a surgeon and a general at the same time. (St
Aubyn The Patrick Melrose Novels: Never Mind, Bad News, Some Hope and
Mother’s Milk, 2012)
Still, the frequency of the word nicotine (64) might be interesting in this
context, since this is the term that tends to be used in reports on the medical
aspects of smoking, sometimes even in a forensic framework. Moreover, a
nicotine-free cigarette appears in Philip Kerr’s A Philosophical Investigation:
(12) Jake lit a cigarette, nicotine free, but the smoke felt good in her lungs,
and picked up her PC and inserted Gilmour’s information disk. (Kerr A
Philosophical Investigation, 1992)
References like these illustrate that novels react to changes in the cul-
tural discourse on smoking and are thus embedded in the cultural his-
tory of tobacco consumption. In Philip Kerr’s novel, a techno-thriller
written in 1992 but set in the early twenty-first century, the author
apparently extrapolates future developments from the anti-smoking
campaigns that were well underway in the 1990s.
131
4.1.2 The Cigarette Script in French
Table 2 shows the most productive tobacco-related RLTs in the French

corpus.
Similarities with the English Corpus

The highest frequency ranks correspond to those we identified in the
English part of the corpus: allumer une cigarette (light a cigarette) and
fumer une cigarette (smoke a cigarette). Moreover, these constructions
fulfil the same narrative functions. Lighting a cigarette/allumer une ciga-
rette, for example, is used for describing tension and nervousness experi-
enced by a character (thus fulfilling a descriptive and, more specifically,
an affective discursive function)1:
(13) Lui, là-haut, retrouvant son matelas pneumatique, privé de son

jumeau, allumait nerveusement une cigarette, s’en repentait, l’éteignait
en l’écrasant sur une solive. (Bazin Un feu dévore un autre feu, 1978) (Up
there, getting back on his inflatable mattress, deprived of his twin, he nerv-
ously lit a cigarette, repented of it, extinguished it by crushing it on a beam.)
Moreover, novels tend to suggest that lighting a cigarette promises a

brief moment of tranquillity, a temporary escape from the worries of
the world. This notion, which has been popularized by advertisements
throughout the twentieth century, may at times be presented in a some-
what satirical tone, as the example below illustrates:
(14) Il écoute Natalie Dessay chanter Mozart, il allume une cigarette

blonde, personne ne l’emmerde, personne ne lui parle, ne le touche, ne
lui demande son avis, il n’a aucun avis, ne prend aucune décision, son
portable est éteint. (Olmi Nous étions faits pour être heureux, 2012) (He
listens to Natalie Dessay singing Mozart, he lights a light-tobacco cigarette,
132
F. Grossmann et al.
Table 2 Most frequent RLTs related to the cigarette in the French sub-corpus
RLTs Corpus Contrast Disp. Disp. Log-
freq. LIT corpus corpus Contrast likelihood
(CONT) LIT corpus
freq. (CONT)
Allumer une cigarette 279 17 95 5 526,1910575
(light a cigarette)
Fumer une cigarette 145 33 71 5 188,745451
(smoke a cigarette)
Tirer une bouffée 54 0 33 0 10000
(take a puff)
Rouler une cigarette 44 1 18 1 92,6422481
(roll a cigarette)
Souffler la fumée 29 0 24 0 10000
(blow smoke)
Sortir un paquet de 18 (14 for 0 12 0 10000
cigarettes/de tabac paquet de
(take out a pack of cigarettes )
cigarettes/tobacco)
Sortir une ciga- 15 0 13 0 10000
rette (take out a
cigarette)
Tirer sur son cigare 17 0 11 0 10000
(draw on his cigar)
Chercher des cig- 11 0 11 0 10000
arettes (look for
cigarettes)
nobody gets on his nerves, nobody talks to him or touches him, or asks
for his opinion, he has no opinion, takes no decision, his mobile phone is
switched off.)
Similar to the English corpus, the erotic implications of smoking/pre-

paring to smoke play a minor role in the French corpus. We only found
a few examples of the seduction scenario in the crime sub-corpus:
(15) Elle croise les jambes, allume une cigarette. (Giébel Les morsures de
l’ombre, 2007) (She crosses her legs, lights a cigarette.)
In the sub-corpus of General fiction, it is elegance or charm rather

than eroticism that seems to be associated with the action of lighting a
cigarette:
133
(16) J’ai été frappé par sa beauté et par ses gestes nonchalants pour
allumer une cigarette ou poser à côté son verre d’orangeade dont elle
aspirait le contenu à l’aide d’une paille. (Modiano Dimanches d’août,
1986) (I was struck by her beauty and by her nonchalant gestures while
lighting a cigarette or setting down her glass of orangeade, whose contents
she sucked up with the help of a straw.)
The theme of seduction may even be integrated into a somewhat maca-

bre scenario, as the following example from a novel by Leila Slimani
illustrates (featuring a situation that is admittedly characteristic of this
particular author rather than of the genre in general):
(17) Il est tout à elle. Pour une fois, il ne peut ni s’enfuir, ni refuser la
conversation. Un bras derrière la tête, les jambes croisées, elle allume une
cigarette. Elle se déshabille. Nue, allongée contre le cadavre, elle caresse sa
peau, elle le serre contre elle. Elle pose des baisers sur ses paupières et sur
ses joues creusées. (Slimani Dans le jardin de l’ogre, 2014) (He is all hers.
For once, he can neither escape nor refuse to talk. One arm behind her
head, legs crossed, she lights a cigarette. She undresses. Naked, stretched
out against the corpse, she caresses his skin, she presses him against her-
self. She puts kisses on his eyelids and on his sunken cheeks.)
The collocation fumer une cigarette (smoke a cigarette) is often, like its
English counterpart, an element that appears in the description of an
individual as seen through the eyes of another character:
(18) Quand elle arriva devant le bâtiment préfabriqué, elle vit la silhou-
ette de M. Filippi, près du pilier B. Il était toujours vêtu de son com-
plet bleu-gris, et il fumait une cigarette en regardant devant lui. (Le Clezio
Mondo et autres histoires, 1978) (When she arrived in front of the prefab-
ricated building, she saw M. Filippi’s silhouette, near pillar B. He was still
dressed in his blue-grey suit, and he smoked a cigarette while looking in
front of himself.)
By showing the character smoking while “looking in front of himself ”,

the author evokes a more concrete visual image and simultaneously pro-
vides information on the character’s psychological state.
134
F. Grossmann et al.
Some Unique Features in the French Corpus

Two RLTs seem specific to the French corpus: tirer une bouffée (take a
puff) and souffler la fumée (blow smoke). Yet, although the RLT blow
smoke occurs only 12 times in the English sub-corpus, a manual extrac-
tion shows that the corresponding collocation is also productive in
English, and well dispersed (55 occurrences, with a dispersion of 38).
The expression is often indicative of a lack of consideration for others,
of impolite demeanour, for example when a character blows smoke in
the direction of the interlocutor:
(19) “Erm, well, actually” – I say in a constricted voice and then stop
mid-sentence as he blows smoke in my face. (Potter Me and Mr Darcy, 2007)
We found evidence of a similar use in the French corpus, even if it is rare:
(20) Le Baron approche ses lèvres craquelées de mon visage, et me souffle

doucement la fumée putride de son cigare dans la bouche. Je tousse, suffo-
que, crache. (Tomas Notre-Dame des Loups, 2014) (The Baron brings his
cracked lips close to my face and softly blows the putrid smell of his cigar
into my mouth. I cough, choke, spit it out.)
Describing the attitude of a smoker who is blowing smoke serves to

allude to the character’s mood or his feelings (descriptive and specifically
affective discursive function, see Appendix B):
(21) Elle bascule la tête en arrière, ferme les yeux pour souffler au plafond la
fumée de sa cigarette. Elle flotte dans un monde imprécis, elle tourne la tête
vers lui, le retrouve, lui sourit. Elle finit son verre de vin, écrase son mégot
sur la table. (Chevrier Madame, 2014) (She tilts her head back, closes her
eyes as she blows the smoke of her cigarette at the ceiling. She is floating in a
blurred world, she turns her head towards him, sees him again, smiles at
him. She drains her glass of wine, crushes her cigarette stub on the table.)
In the English corpus, the same function is fulfilled by diverse col-

locations, including the expressions puff of smoke or take a drag of his
cigarette:
135
(22) Miss Brown inhaled a deep puff of smoke, and then exhaled it very
gradually. (Barrie Return to Tremarth, 1969)
(23) “What are you doing Friday?” he persists, raising his eyebrows and
taking a drag of his cigarette. (Potter Me and Mr Darcy, 2007)
In the French corpus, tirer une bouffée (take a puff) is used to provide
details on the smoker’s personal ritual, referring to a way of either enjoy-
ing the moment (example 24) or impressing others (example 25):
(24) Pour se calmer tout à fait, il avait besoin de fumer. Il sortit de la

voiture, verrouilla la portière et fit quelques pas sur le trottoir avant d’al-
lumer une cigarette et d’en tirer une longue bouffée. (Musso Demain, 2013)
(In order to calm down completely he needed to smoke. He left the car,
locked the door and took a few steps on the pavement before lighting a
cigarette and taking a long drag.)
(25) Elle s’avança lentement. Ses pas semblaient étudiés pour faire
apprécier la perfection de sa longue silhouette perchée sur des talons
aiguilles. Iris nota avec amusement que rien ne manquait à sa panoplie
de femme fatale puisqu’elle n’avait pas oublié un long fume-cigarette
dont elle tira une bouffée en s’arrêtant devant la jeune fille (Lambert Au-
delà des ténèbres, 1981) (She advanced slowly. Her steps seemed studied
for the appreciation of the perfection of her long silhouette, cantilevered
on stiletto heels. Iris noted with amusement that nothing was missing in
her arsenal of the femme fatale since she hadn’t forgotten a long cigarette
holder from which she took a drag while stopping in front of the girl.)
Where brand names are concerned, it comes as no surprise that more

Gauloises are smoked in French novels (35 occurrences) at the expense
of Gitanes (a mere 6 occurrences). Other brands mentioned are 6
Camel; 5 Craven; 3 Dunhill; 2 Winston; 4 Marlboro; 4 Davidoff (1 cig-
arette and 3 cigars). (In French novels, the term Chesterfield refers to a
leather sofa rather than to a cigarette brand.) In the French corpus, the
most commonly used colloquial term for cigarettes is clope (93 occur-
rences). 57 of these occurrences occur in crime novels, where they help
to establish a casual tone.
136
F. Grossmann et al.
4.2 The Drink Script in English and French
Methodologically, extracting the most productive RLTs related to alco-

hol from the English and the French corpus turned out to be quite dif-
ficult since the most frequent expressions in the drink script may also
refer to non-alcoholic beverages. Thus, we had to disambiguate the data
manually, examining the expressions in their context.
4.2.1 The Drink Script in English
Table 3 shows the frequency of have a drink in English, which turns

out to be by far the most frequent verbal collocation used to express the
consumption of alcohol.
Although the expression have a drink remains under-specified in
some cases, in its prototypical use it clearly refers to the consumption of
alcohol. The two examples below, however, illustrate the possible ambi-
guity of have a proper drink:
(26) Laura sighed deeply “To hell with tea, let’s have a proper drink.” Will
produced a bottle of whisky. (Fforde Practically Perfect, 2006)
(27) I haven’t had a proper drink since me and Brubeck got some water
from a tap in the church, and the rules say that you can’t knock on a door
and ask for a glass of water in a town the way you can in the middle of
nowhere. (Mitchell The Bone Clocks, 2014)
In the vast majority of its uses in novels, have a drink correlates with the
description of a relaxing situation, which can involve others or only one
character. We have identified several recurring patterns (or motifs, see
the third part of Chapter 1 by Novakova and Siepmann in this volume)
that clarify the meaning of the main collocation:
a. Invitations (“X offers a drink to Y and/or other people”)
Invitations are by far the most frequent context for have a drink (93
occurrences).
137
Table 3 Most frequent RLTs related to the consumption of alcohol in the

English sub-corpus
RLTs Frequency Frequency Dispersion Dispersion Log-
in corpus LIT in the in corpus in Contrast likelihood
Contrast LIT corpus
corpus (CONT)
(CONT)
Have a drink 249 (237 66 83 4 360,0613
after
sorting)
Go for a 58 15 34 2 84,8027824
drink
Buy a drink 52 11 41 1 82,57863223
Put the glass 42 0 35 0 10000
Come for a 34 1 24 1 79,062325
drink
Ask for a 32 2 25 1 68,43081751
drink
Reach for the 30 1 25 1 69,01495155
bottle
Take a sip of 30 0 20 0 10000
wine
Pour a glass 27 1 22 1 61,50166
of wine
Have a glass 27 4 19 1 48,22746836
of wine
Drink a cup 21 0 19 0 30212347
Pour a 22 (20 after 0 17 0 10000
measure sorting)
(28) Why don’t you have a drink with us? said Cass. We’ve got plenty.
(Garnett Bikini Planet, 2000)
(29) Look, have you got to go straight home? Will they worry where
you’ve got to? I’ve just had three cups of Scotch. I think I need more.
Come and have a drink with me. (Swift Waterland, 1983)
(30) No bother at all. Have a drink, sherry, whisky, gin? There’s some
Campari somewhere. (Murdoch The Good Apprentice, 1985)
The invitation can be in the form of a direct question or an explicit

request (Will you have a drink? Come and have a drink, Let’s just have a
138
F. Grossmann et al.
drink on + date). It can also be phrased more indirectly (e.g., I presume

you can have a drink; You should absolutely have a drink ). This finding
suggests that in novels the consumption of alcohol appears, first and
foremost, in the context of social occasions, typically before a discussion
or other forms of interaction.
b. Requests (“Y asks X for a drink”)
The “drink request” is much rarer than the offer, but sometimes a char-
acter apparently feels the need to ask for a strong drink, for instance to
imbibe courage:
(31) “Before we go, could I possibly have a drink? – A serious drink? I

can’t think why, but I suddenly feel like one.” The ante-room was
deserted. Alexander poured two glasses of Johnnie Walker Black Label
and made out the chit. (Amis Russian Hide-and-Seek, 1980)
c. Descriptive/narrative forms with a temporal indication

In such uses, the reference to the consumption of alcohol is integrated
into the narration (diegetic function) and does not appear in the
dialogue:
(32) Without asking, he poured a brandy for me. When I had arrived in
England, I had never had a drink of hard liquor. As I took the glass from
his hand, I realized how much I had changed. (Simonsen Searching for
Pemberley, 2007)
(33) I passed the pub where I’d first had a drink with Carla only a week
earlier and, after hesitating for a moment, went inside. (Kernick The
Business of Dying, 2002)
The expression go for a drink (often combined with out ), which is

much less frequent but still relatively productive (58 occurrences), typ-
ically has a diegetic function and introduces a change of place. We also
encounter the same functions as those found in dialogues with have a
drink, albeit in different proportions (only a dozen occurrences for the
direct invitation with go for a drink ).
139
4.2.2 The Drink Script in French
Table 4 shows that the two most frequent collocations in French concerning
our topic are boire un verre (have a glass) and prendre un verre (take a glass),
according to the RLTs extracted. As before, we were careful to exclude irrele-
vant occurrences (e.g., boire un verre d’eau/drink a glass of water).
Here we have far fewer forms expressing a direct invitation than in the
English corpus (only 18 occurences with boire un verre [drink a glass], out
of a total of 196), but there are more occurrences of direct invitations with
prendre un verre (take a glass, 23 occurrences out of a total of 96). Overall,
however, there is a discrepancy in this respect between what we observed
in the French and the English corpora. A possible explanation is that the
invitation to have a drink appears less often in dialogues and is generally
more integrated into the narrative, as the example below illustrates:
(34) N’ayant toujours pas envie de rentrer, j’acceptai sa proposition d’aller

prendre un dernier verre au Grand Café, ouvert toute la nuit, dans le quar-
tier de l’Opéra. (Abécassis Les tribulations d’une jeune divorcée, 2005) (Still
not wanting to go home, I accepted his suggestion to go and have a last
glass in the Grand Café, which is open all night, in the Opéra district.)
Although the expressions formed with the verb siroter (sip) are not
among the extracted RLTs (due to our very strict criteria), this verb is
still very common in referring to the consumption of alcohol. It is par-
ticularly so in French, for implying a particular way of consumption
suggestive of waiting as well as recklessness.
Table 4 shows the main types of alcohol represented in the English
and French corpora. We took care to group the different specifications
of wine together; that is, for example Chardonnay and Beaujolais were
subsumed under the category wine. Similarly, we assigned beer and ale
as well as whisky and scotch to the same category (Fig. 1).
For this graph, we have chosen a minimum threshold of 3% for
at least one of the sub-corpora (English or French), except for the
generic term alcohol, which we considered separately and therefore
omitted from the table. Although this generic term does not refer to
140
F. Grossmann et al.
Table 4 Most frequent RLTs related to the consumption of alcohol in the French
sub-corpus
RLTs Freq. cor- Freq. Disp. cor- Disp. Log-
pus LIT Contrast pus LIT Contrast likelihood
corpus corpus
(CONT) (CONT)
Boire un verre 223 (196 144 113 5 131,8958319
(drink a glass) after
sorting)
Prendre un verre 103 (96 54 69 6 76,40906935
(take a glass) after
sorting)
Boire du vin (drink 83 28 56 5 87,27337182
wine)
(Se) servir un verre 103 (89 8 61 2 188,2784155
(serve [oneself] a after
glass) sorting)
Remplir un verre 35 (31 0 29 0 10000
(top up a glass) after
sorting)
Servir un verre de 18 0 17 0 10000
vin (serve a glass
of wine)
Boire une gorgée 17 0 15 0 10000
de vin (drink a
mouthful of wine)
S’emparer du verre 16 0 10 0 10000
(seize the glass)
S’emparer de la 14 0 11 0 10000
bouteille (seize the
bottle)
Remplir les coupes 14 0 10 0 10000
(top up the
goblets)
Tendre un verre 34 (6 after 1 29 1 70,08891624
(offer a glass) sorting)
a specific type of beverage, how it is used in the corpus is instructive.

The other alcoholic beverages we agglomerated in the column “others”
and describe in more detail below (see Fig. 2). Some of the results are
quite surprising, for instance the fact that beer appears more frequently
in the French corpus, whereas wine is equally common in both cor-
pora. By contrast, it is not surprising that more sherry is consumed in
141
ϰϱй ϰϭй ϰϭй

ϰϬй
ϰϬй
ϯϱй
ϯϬй
Ϯϱй ϮϮй
ϮϬй
ϭϱй ϭϮй
ϭϬй ϳй ϳй
ϱй ϰй ϯй
ϱй ϯй ϯй ϯй ϯй ϯй
ϭй ϭй ϭй Ϭй Ϭй
Ϭй
ŶŐůŝƐŚ &ƌĞŶĐŚ
Fig. 1 Most frequent types of alcohol in the English and French sub-corpora
(n English = 8046; n French = 6837)
ϰϱй
ϰϭй
ϰϬй
ϯϱй ϯϯй
Ϯϵй
ϯϬй
Ϯϲй
Ϯϱй ϮϮй Ϯϭй
ϮϬй ϭϳй
ϭϱй ϭϮй
ϭϬй
ϱй
Ϭй
ŽƵƌďŽŶ ŚĂŵƉĂŐŶĞ ŝĚĞƌ dĞƋƵŝůĂ
ŶŐůŝƐŚ &ƌĞŶĐŚ
Fig. 2 “Other” types of alcohol in the English and French sub-corpora, n = 229
in the English corpus, n = 239 in the French corpus
English-speaking countries. The slight over-representation of rum in the

English corpus can partly be explained by the fact that this type of alco-
hol enters more into the preparation of cakes—at least according to the
novels in our corpus.
142
F. Grossmann et al.
We found that the differences between French and English are not
significant for “other” types of alcohol—with the exception of bourbon,
which is (somewhat counterintuitively) more frequent in the French
corpus, and, to a lesser extent, tequila, which has a few more occur-
rences in the English corpus.
Although there are proportionally fewer terms referring to different
types of alcoholic beverages in the French than in the English corpus,
the generic term alcohol occurs more frequently in French, which speaks
for a closer look at how this generic term is used as being worthwhile.
Firstly, it may co-occur with a term specifying what type of alcoholic
beverage is meant (in anaphoric or cataphoric use):
(35) Des boîtes de conserve qui ont roulé sous la table. Une bouteille de
whisky éclatée dont le liquide blond s’est réfugié dans un coin, diffusant
son odeur d’alcool tout autour. (Adam Peine perdue, 2014) (Cans which
have rolled under the table. A smashed bottle of whiskey whose fair liquid
has escaped into a corner, spreading its alcoholic vapours all around.)
Secondly, in contexts where the term alcohol is used without any fur-
ther specification, it typically serves to portray addiction or personal
decay, or it appears in a medical context, especially in crime fiction and
romance novels:
(36) – Vous traitez toutes les dépendances, en fait? – Oui. On a beaucoup

de « polytoxicos » : dérèglements alimentaires, alcool, cocaïne, la totale.
(Alexis Je n’irai pas chez le psy pour ce con, 2009) (– Do you really treat all
addictions? – Yes. We have many polydrug abusers: eating disorders, alco-
hol, cocaine, the lot.)
(37) Donc vous ne pouvez rien me dire sur ses habitudes. Femmes, mœurs
sexuelles, drogue, alcool? (Arnaud Pierres de sang, 1999) (That means you can-
not tell me anything about his habits. Women, sexual morals, drugs, alcohol?)
We found the same two types of usage in English, even if with fewer
occurrences. This may suggest that French writers tend to prefer gen-
eralizing the experience of the characters in sociological terms, while
English writers are apparently more inclined to making the references
concrete.
143
5 Comparison of Two Linguistic Patterns

(or Motifs): Take a Sip of and Boire une
Gorgée de
When it comes to analyzing a specific pattern, it is interesting to com-
pare not only languages but also different genres. In Fig. 3, we show the
results for the distribution of the expression take a sip across the differ-
ent genres in the English corpus.
It is striking how much more common this pattern is in romance
novels (abbreviated as ROM) than in the other genres. It suggests an
affinity between take a sip and the “romantic universe” typically created
in this genre. A closer look at the examples, however, reveals that alco-
holic beverages are not necessarily what is consumed by the characters
in this genre. In fact, we find roughly the same statistical distribution
for the expression take a sip used with all kinds of beverages (includ-
ing coffee, tea, water, etc.) before sorting and after sorting for the same
expression when used with terms denoting alcoholic beverages (wine,
beer, whisky, vodka, etc.). These findings suggest that it is the manner
of drinking expressed by the motif take a sip rather than any particular
ϱϬй
ĞĨŽƌĞƐŽƌƚ͘ ŌĞƌƐŽƌƚ͘
ϰϰй
ϰϱй
ϰϭй
ϰϬй
ϯϱй
ϯϬй
Ϯϱй
ϮϬй ϭϴй ϭϳй ϭϳй

ϭϱй
ϭϱй
ϭϬй ϭϭй
ϭϬй ϴй ϳй
ϲй
ϰй
ϱй
Ϭй
Z/D ,/^d 'E ZKD ^& &z
Fig. 3 Take a sip in the English corpus before and after sorting, n = 303 before
sorting; 202 after sorting
144
F. Grossmann et al.
type of drink that is characteristic of English romance novels—which

empirically supports the assumption that certain demeanours on the
part of the characters may be deemed typical of a particular literary
genre. In French, siroter (to sip) when followed by a noun referring to
alcohol plays a similar role and is in fact well represented in romances.
The distribution of boire une gorgée in the French corpus is somewhat
less clear (see Fig. 4). This expression is not a straightforward equivalent
of take a sip, corresponding rather more to drink a mouthful, but it is
presumably still the closest extant equivalent.
In French, the collocation is more frequent in two genres (romance
novels and crime fiction), but the overall distribution is clearly more
balanced than in English. This suggests that the collocation boire une
gorgée is used more widely than take a sip. It is also less marked than its
English near-equivalent, perhaps because of being less associated with
the connotations of refinement, decorum and tasteful pleasure that pro-
mote the use of take a sip in a romantic context.
We can check this assumption by comparing the RLT take a sip
of wine with its near-equivalent boire une gorgée de vin in the two
ϯϱй
ϯϬй Ϯϵй
Ϯϳй
Ϯϲй
Ϯϱй
ϮϬй ϭϵй
ϭϳй ϭϳй
ϭϱй ϭϰй
ϭϯй ϭϯй
ϭϭй
ϭϬй
ϳй ϳй
ϱй
Ϭй
Z/D ,/^d 'E ZKD ^& &z
ĞĨŽƌĞƐŽƌƚ͘ ŌĞƌƐŽƌƚ͘
Fig. 4 Boire une gorgée in the French corpus before and after sorting, n = 207
before sorting; 113 after sorting
145
sub-corpora. In English, we found 30 occurrences of this collocation

(with a dispersion of 20). In French, boire une gorgée de vin appears
only 17 times, with a dispersion of 15, but two very similar expressions
(prendre une gorgée de vin and avaler une gorgée de vin ) appear 7 and 6
times, respectively. When considering only the nominal collocation, we
found 56 occurrences of gorgée de vin and 44 of sip of wine. The distri-
bution among literary genres is quite different in English and French
and confirms our findings for the verbal collocations take a sip and boire
une gorgée:
As Fig. 5 illustrates, sip of wine is used much more frequently in the
romance novels in our corpus (ROM) than its French equivalent gorgée
de vin. Even more than in the case of the verbal collocation, we found a
far more balanced distribution in French, except for science fiction and
general fiction, where the collocation appears only rarely.
Conversely, it is remarkable that the expression sip of wine sees very
little use in English fantasy fiction, whereas its near-equivalent gorgée de
vin is found frequently in French fantasy. The difference between the
English and the French data observed for the fantasy category can be
ϱϬй
ϰϱй
ϰϱй
ϰϬй
ϯϱй
ϯϬй
Ϯϰй
Ϯϱй
Ϯϭй Ϯϭй Ϯϭй
ϮϬй ϭϴй
ϭϲй
ϭϱй
ϭϭй
ϵй
ϭϬй
ϳй
ϱй
ϱй
Ϯй
Ϭй
ZKD Z/D ,/^d 'E ^& &z
^ŝƉŽĨǁŝŶĞ 'ŽƌŐĠĞĚĞǀŝŶ
Fig. 5 Sip of wine and gorgée de vin in the English and French corpora
146
F. Grossmann et al.
explained lexically. After all, sip of wine implies a refined behaviour,

while gorgée de vin seems more compatible with a “manly” demeanour
that better suits characters in fantasy fiction.
Our qualitative study shows that, despite a certain imbalance in
terms of distribution, the expressions have roughly the same functions
in both languages:
a. The creation of referential micro-universes characteristic of each of
the literary genres
In the romance novels, for example, lovers live their passion in idyllic
landscapes (see example 38) or in luxurious homes, they share alcoholic
beverages during the night in front of a cosy fire, and wine becomes the
symbol of their union (example 39):
(38) The blond wood tables sat to one side of the huge window that
looked out over Dún Laoghaire harbour, so they could sip a glass of wine,
talk about their days and relax with the exquisite view in the background.
(Kelly It Started with Paris, 2014)
(39) “Je prends une gorgée de vin et je la lui verse dans la bouche. Elle
prend une gorgée de vin et elle me la verse dans la bouche. Les gorgées
deviennent de plus en plus grosses et les bouches s’ouvrent de plus en plus
grand.” (Labrèche Borderline, 2000) (I take a mouthful of wine and pour
it into her mouth. She takes a mouthful of wine and she pours it into my
mouth. The mouthfuls become bigger and bigger and the mouths open
more and more.)
Example 39 illustrates a homosexual love scene, in which, as Boisclair

(2012, 91) points out, “desire is reciprocal, positions are interchangeable”—
something that is aptly expressed by the drinking script.
b. Textual functions which are not specific to a literary genre
Take a sip of wine and boire une gorgée de vin are mainly used in the con-
text of dialogues (either announcing or interrupting them): typically, a
character takes a sip of wine before speaking. This action often suggests
a pause before saying something difficult or embarrassing. Thus, nov-
elists draw upon this motif to flesh out the depiction of psychological
147
states (cognitive discursive function, see Appendix B), but also to alter
the routine of turn-taking by describing the context of speaking (infra-
narrative discursive function, see Appendix B). The manner of holding
a drink and taking a sip of wine sometimes also serves to describe the
characters’ body language, and thus their attitudes and state of mind
(indirectly descriptive function):
(40) Manin enfourna une fourchette de pâtes dans sa bouche et prit le

temps de bien les mâcher. «Je crois que tu as raison mais nous n’en sau-
rons plus que si nous parvenons à nous entretenir avec lui.» Il but une
gorgée de vin. «Tu as remarqué qu’il évoquait aussi une lumière blanche
après un long tunnel et un vol dans les airs?» (DOA La ligne de sang,
2004) (Manin shoveled a forkful of pasta into his mouth and took the
time to chew it well. “I believe you are right, but we will know more
about it if we manage to talk with him.” He drank a mouthful of wine.
“You have noticed that he also remembered a white light at the end of a
long tunnel and being airborne?”)
(41) Angelica took a glass from her and sat down, not quite sure how
to play this unusual turn of events. She took a tentative sip of wine; she
wasn’t a great drinker, and when she did, she favoured sugary alcopops.
She shuddered slightly as the sharp tang of gooseberry hit her taste buds.
“So,” she ventured finally. (Henry The Long Weekend, 2012)
In example (40), the character drinks a sip of wine before speaking

again. In example (41), the author focuses more on the psychological
description: here, the way of drinking correlates with embarrassment.
6 Conclusion
Our study confirms that references to the different types of alcoholic
beverages and tobacco, as well as to the ways of consuming them, are
mainly used by writers to allude to the characters’ attitudes and emo-
tions; in other words, details about the consumption of alcohol and
tobacco are primarily a means of (implicit) characterization of both
main and marginal personae. On the whole, the diegetic function seems
148
F. Grossmann et al.
less important, but this may be due to the fact that we have not system-
atically studied how the places where characters drink or smoke (bars,
restaurants, etc.) are described; the wider textual contexts of the ciga-
rette and drink scripts certainly deserve further investigation. Showing
characters to be drinking or smoking is also a means of announcing or
suspending dialogue or action. This means these passages may signifi-
cantly impact the pace of the narrative.
Our corpus study thus clearly reveals that the scripts related to the
consumption of alcohol and tobacco primarily serve a dual function.
The first one depends directly on the characteristics of the literary
genre. The referential universe specific to a literary genre—which can
vary according to the linguistic and cultural background—determines
the frequency of one or another of the collocations associated with a
script. Worth noting is that, overall, the RLTs found in the English cor-
pus and those identified in the French one overlap, although obviously
the details differ. What varies are not so much their referential or textual
functions (which, in fact, turn out to be pretty similar) but their fre-
quency within one and the same genre. This we amply demonstrated
with the case study focusing on take a sip of wine: as a complex colloca-
tion, it is characteristic of romance novels in English, while this is not
the case for its French near-equivalent boire une gorgée de vin.
The literary genre, rooted in a given culture and history and tend-
ing to be informed by specific literary traditions, is certainly a decisive
factor here. Still, it must also be acknowledged that there is no seman-
tic one-to-one correspondence between the two expressions take a sip
of wine and boire une gorgée de vin, which differ to a certain extent at
least in terms of their connotations. Of course, to proceed further in
the comparison, we should look at the whole semantic field concerned:
for example, siroter (to sip), whose meaning is very close to that of take
a sip, which is often used in French. This is one of the limits of our
RLTs-based approach, which, while it is a good starting point, eventu-
ally must be supplemented by examining less frequent or less typical
expressions.
The second function transcends literary genres, since forms of human
interaction like smoking a cigarette or drinking an alcoholic beverage
span all genres. This might also go a long way towards explaining why
149
there are no genres, including even the most recent output and science
fiction (where one is presumably most likely to expect radically dif-
ferent types of interaction), that avoid mentioning tobacco or alcohol
consumption altogether. Referencing these addictions remains a power-
ful descriptive and narrative resource for portraying human beings and
their interactions, even if it remains to be seen how they might evolve in
the works of the next generations of novelists.
Acknowledgements We are grateful to Laetitia Gonon and Iva Novakova for

their thorough reading of this chapter and their helpful suggestions.
References
Bentolila, Eric. 2016. “Le roman policier français de 1970 et 2000: une ana-
lyse littéraire.” PhD diss., Université Grenoble Alpes.
Boisclair, Isabelle. 2012. “Mourir ou jouir: détournement et retournement des
scripts sexuels hétéronormatifs dans Borderline de Marie-Sissi Labrèche.”
Francofonia 62: 83–95.
Burns, Eric. 2006. The Smoke of the Gods: A Social History of Tobacco.
Philadelphia: Temple University Press.
Ciobica, Irina, Alin Ciobica, Daniel Timofte, and Stefan Colibaba. 2015.
“James Joyce and Alcoholism.” International Letters of Social and Humanistic
Sciences 59: 146–56.
Eder, Jens, Fotis Jannidis, and Ralf Schneider. 2010. “Characters in Fictional
Worlds: An Introduction.” In Characters in Fictional Worlds: Understanding
Imaginary Beings in Literature, Film, and Other Media, edited by Jens Eder,
Fotis Jannidis, and Ralf Schneider, 3–64. Berlin and New York: De Gruyter.
Emanuel, Michelle. 2005. “Smoking Guns and Lingering Pipes: Tobacco
Imagery in the Novels of Léo Malet.” Romance Notes 46 (1): 87–95.
Genette, Gérard. 1972. Figures III. Paris: Seuil.
Gonon, Laetitia, Vannina Goossens, and Iva Novakova. 2019. “Les phraséolo-
gismes spécifiques à deux sous-genres de la paralittérature: le roman policier
et le roman sentimental.” In Actes du colloque Phraséologie française, 21–22
September 2017. Arras: Université d’Artois.
Grylls, David. 2006. “Smoke Signals: The Sexual Semiotics of Smoking in
Victorian Fiction.” Journal of the English Association 55 (211): 15–35.
150
F. Grossmann et al.
Harp, Richard. 1988. “Tobacco and Raymond Chandler.” Clues: A Journal of

Detection 9 (2): 95–104.
Harvey, A. D. 2014. “The Cigarette as Style Accessory.” Critical Quarterly 56
(1): 83–87.
Herman, David. 2006. “Narrative: Cognitive Approaches.” In Encyclopedia of
Language and Linguistics, edited by Keith Brown et al., 2nd ed., 452–59,
section edited by Catherine Emmott. Oxford: Elsevier.
Herman, David. 2009. Basic Elements of Narrative. Oxford: Wiley-Blackwell.
Herman, David. 2013. “Cognitive Narratology.” In The Living Handbook of
Narratology, edited by Peter Hühn et al. https://www.lhn.uni-hamburg.de/.
Kalay, Faruk. 2016. “A Humoristic and Satirical Perspective to Jewish Classes
in Goodbye, Colombus by Philip Roth.” Toplum ve Bilim/Science & Society
5 (8): 3246–58.
Lacroix, Alexandre. 2001. Se noyer dans l’alcool. Paris: Presses Universitaires de
France.
Levet, Natacha. 2012. Sherlock Holmes, De Baker Street au grand écran. Paris:
Editions Autrement.
Montémont, Véronique. 2009. “Queneau, Perec, Duras: Trois manières de
boire dans le roman français.” COnTEXTES. http://journals.openedition.
org/contextes/4525.
Muller, Philippe. 2013. “Cet obscur objet de désir: à propos de la construc-
tion want GN et de ses gloses.” Corela. http://journals.openedition.org/
corela/2997.
Onwordi, Toni Kan. 2003. “Cigarettes as Characters in Heinrich Böll’s Short
Fiction.” Heinrich Böll Stiftung, 10 January 2008. https://www.boell.de/en/
content/cigarettes-characters-heinrich-boells-short-fiction.
Riffaterre, Michael. 1986. “On the Diegetic Functions of the Descriptive.”
Style 20: 281–94.
Rozenberg, Félix. 2015. “L’Alcool, un ami et un ennemi. Le rôle de l’al-
cool dans Les particules élémentaires et La carte et le territoire de Michel
Houellebecq.” Mémoire de master, Lunds Universitet, Lunds.
Schank, Roger C., and Robert P. Abelson. 1977. Scripts, Plans, Goals, and
Understanding. Hillsdale, NJ: Lawrence Erlbaum Associates.
Stevenson, Robert Louis. 1979. Dr Jekyll and Mr Hyde and Other Stories.
Harmondsworth: Penguin.
6
French and American Science Fiction
During the Nineties: A Contrastive Study
of Fiction Words and Phraseology
Laetitia Gonon and Olivier Kraif
1 Introduction
For several years now, the members of the PhraseoRom research pro-
ject have worked on recurrent lexico-syntactic constructions in con-
temporary novels to demonstrate the specificity of these constructions
in literary language in general and in literary subgenres in particular.
But not only is the subgenre of science fiction distinguished by spe-
cialized phrases and motifs common to the novels of this literary cate-
gory, but also by neologisms often specific to individual authors. This
poses a number of questions: How can we study what seems unique
if our methodology is based on extracting quantitatively significant
occurrences from large digitized corpora? Also, is it possible to study
L. Gonon (*) · O. Kraif

O. Kraif
e-mail: olivier.kraif@univ-grenoble-alpes.fr
https://doi.org/10.1007/978-3-030-23744-8_6
152
L. Gonon and O. Kraif
an individual author’s neologisms even if our methodology is geared

instead to showing the links between several authors of the corpus?
These are germane questions, seeing that the PhraseoRom project
aims, among others, to make better identification of fictional gen-
res possible. The critics agree in fact that neologisms are among sci-
ence fiction’s (SF’s) defining features; these refer to realities unfamiliar
to readers, such as futuristic inventions, advanced technology, anthro-
pomorphic species from other planets, extraterrestrial fauna and flora,
and so on, that have no referents in reality. Angenot calls them “fiction
words” (mots-fictions; 1978, 81). Such words coined by novelists—and
their imaginary referents—then constitute what is called a xenoency-
clopedia (Saint-Gelais 1999, 140). As elements of this “factual report-
ing of fiction” (Suvin 1979, 62), fiction words are at the core of what
Suvin calls “cognitive estrangement” (ibid., 60), a kind of distancing
with respect to both reality and fiction that melds a sense of wonder and
rationality under conditions of plausibility.
Science fiction seems to be a predominantly Anglo-Saxon genre:
“Malgré les nombreuses et riches traditions non anglophones d’écrit-
ure de science-fiction, un nombre élevé de tropes, de motifs, d’icônes et
de modèles historiques dans le mégatexte de la science-fiction provient
de, ou a été élaboré de façon décisive à partir du fonds de la science-fic-
tion anglo-américaine” (Csicsery-Ronay 2013, §15). (Despite the various
and rich non-English science fiction writing traditions, a large number of
tropes, motifs, icons, and historical models in the science fiction megatext
have come—or have been decisively developed from—the fund of Anglo-
American science fiction; our translation.) Has this cultural predominance
influenced the elements of xenoencyclopedias in other literatures? Are
there different ways of using fiction words in languages like French? To
explore the notion of the xenoencyclopedia and its implications through
the study of phraseology, we have adopted a contrasting perspective for this
paper of comparing fiction words in American and French science fiction.
Since the xenoencyclopedia provides narrative background, settings,
and atmosphere (and thus is not necessarily linked to the main novelis-
tic themes), it may surface interesting differences between English and
French science fiction beyond what more traditional approaches centering
on thematic perspectives have revealed. Linguistic material (phraseology,
6 French and American Science Fiction During the Nineties …
153
the lexical formation of neologisms) may vary significantly. The objective

is therefore to study contrastive features relevant at a linguistic level by
drawing upon methods of corpus linguistics and textometry.
Section 2 of the article is dedicated to the description of corpus data
and methodology used to extract two comparable lists of fiction words.
The third section takes up the morphological and semantic properties of
the selected words. In the fourth section, we attempt to delineate par-
ticular properties at a phraseological level by dealing with the syntactic
contexts of fiction words.
2 Corpus and Methodology

2.1 Corpus
The corpus1 consists of 98 science fiction novels. To provide a reliable

contrastive analysis and neutralize diachronic aspects, we limited the
temporal publication span of the novels analysed to texts published in
the 1990s (1990–1999) (Table 1).
The corpus contains data that is comparable in terms of the diversity of
both authors and novels, which satisfies the two most important criteria
for the study. Another key criterion for the selection was the inclusion
of authors widely recognised in the SF genre. When it comes to word
counts, however, the two corpora differ substantially—SF novels written
in English, at least those included in the present corpus, simply are longer
than ones written in French. This is not necessarily a general tendency, of
course: it may just be consequent on the application of our criteria.
Table 1 Quantitative description of the corpora

Authors Novels Words
English (Anglo-American) 25 49 9,000,961
French 21 49 4,361,176
1The corpus was composed by Judith Chambre; she brought together the various English and
French titles of science fiction novels from the 1990s by taking into account the diversity of sub-
genres and the representativeness of the selected books in the field of science fiction.
154
2.2 Method
The tool we used here was Lexicoscope (Kraif and Diwersy 2012),
which has been used in connection with other PhraseoRom pub-
lications. As previous studies (Chambre and Kraif 2017; Kraif and
Sorba 2018) demonstrated, recurring lexico-syntactic trees (RLTs, see
Chapter 1 by Novakova and Siepmann, Sect. 2, in this volume) have
yielded interesting clues for distinguishing between various subgenres.
For instance, recurring trees underlying expressions such as <donner
l’ordre> (give the order), <prendre le pouvoir> (take the power), <le plus
puissant> (the most powerful), <commettre une erreur> (make a mistake),
<dans toutes les directions> (in all directions) appear to be, somewhat
unexpectedly, efficient markers for distinguishing SF from the CRIM
and GEN sub-corpora.
For this paper, we adopted the same corpus-driven approach fol-
lowed in earlier articles and ones written for this volume. Nevertheless,
because words making up the xenoencyclopedia are not found in dic-
tionaries, Lexicoscope cannot label them. They are usually more or less
specific to one author, even to a particular novel. To study the phra-
seology of SF through the xenoencyclopedia, a new heuristic for iden-
tifying these unknown words and their contexts has to be established.
This takes a combination of computer science, corpus linguistics and
stylistic analysis skills. Because xenoencyclopedian words are nearly all
unique, this seems a daunting if not impossible task. However, there is
an out: we can assume that these fiction words appear in a specific con-
text, which exposes their definition to readers without breaking the nar-
rative flow—in other words, the extradiegetic narrator in an SF novel
will not normally intervene to insert explanations. Instead, he or she is
likely to have a character deliver the explanation or limit explanations
to single expressions to avoid slowing down the diegesis. It can there-
fore be assumed that fiction words will generally be preceded or fol-
lowed by certain definitional clues and that this defining context will be
motivated by the plot. Thus, for example, a character not in the know
(as stand-in for the reader) solicits it, parentheses enclose it, or a hyper-
nym (like weapon, creature… ) or an anaphora more generally refor-
mulate it. In effect, fiction words often appear in “didactic segments”
155
(Saint-Gelais 1999; Langlet 2006, 43). These hunches call for statistical
proof. To provide it, in this paper we rely on two types of automatic
RLT extraction.
The extraction is based on a re-annotation of the corpus that takes
unknown words into consideration. These will have been identified
automatically and had their lemmas replaced by the string <FW> (to
denote Fiction Words). This identification would have been achieved by
drawing on a fusion of general-purpose word lists extracted from vari-
ous corpora:
– for English: English Wacky corpus,2 American National Corpus,3

PhraseoRom CONT corpus (see Appendix A) completed by the
Moby Word List available on Project Gutenberg.4
– for French: PhraseoRom CONT corpus completed by the ABU
inflected form dictionary.5
In an effort to extract only general fiction words and avoid capturing

named entities, we marked as <FW> only strings that did not begin
with a capital letter.
By the time the re-annotation phase ended, we had marked 32,698
and 15,149 words, respectively, in the English and French corpora. This
unsurprising difference stems from the English corpus being twice as
large as the French one. In any event, these statistics are not really compa-
rable as the word lists were established in dissimilar ways: the French list
includes the extensive ABU dictionary of inflected forms with 300,000
clean entries, while in English we lacked access to such a resource.
We then extracted RLTs containing unknown words (<FW>) from
the corpora. This made it possible to better identify and analyse recur-
rent lexico-syntactic constructions including a neologism in a narrow
context.
2 http://wacky.sslmit.unibo.it/lib/exe/fetch.php?media=frequency_lists:sorted.uk.word.
unigrams.7z.
3http://www.anc.org/SecondRelease/data/ANC-all-count.txt.
4http://www.gutenberg.org/etext/3201.
5http://abu.cnam.fr/DICO/mots-communs.html.
156
3 The Formation of Fiction Words

Before turning to an analysis of the syntactic contexts of a sample of
fiction words, we deem it useful to first characterize their lexical prop-
erties. Two aspects appear to be relevant: how fiction words are formed
and to which semantic class they belong.
3.1 Methodology
In lieu of studying all the occurrences marked <FW>, we took a ran-

dom sample numbering 500 occurrences in each corpus distributed
over all the texts. These occurrences were then filtered to eliminate can-
didates that were not genuine fiction words, that is: words that were
not on our lists but attested to by at least one reference dictionary,6 for
example:
(1) she subvocalized the half obscenity. (Brin Earth, 1990)
Words with an attested graphic variant in a dictionary, for example: for

booby-trapped one finds
(2) they couldn’t have just boobytrapped the building. (Haldeman Forever
Peace, 1997)
Words corresponding to occasional lexical creations, which could be

called occasionalisms (Poix 2018): they are not intended to create neol-
ogisms within the novel since they do not perform a precise referential
function in the fictional world created by the author. For example, one
may find these sorts of onomatopoeia-like expressions:
(3) Jane played some Thai pop music, cheerful energetic bonging and
strumming. (Sterling Heavy Weather, 1990)
6Larousse online, Petit Robert and Wiktionnaire for French, Collins English Dictionary, Oxford
English Dictionary and Wiktionary for English.

157
Or even occasional compositions or derivations:
(4) the ultravampish look of a bad actress. (Sterling Heavy Weather, 1990)
(5) I still had the hope in my heart, now dwindlingly faint …. (Pohl The
Far Shore of Time, 1999)
However, we did retain words attested by dictionaries in the following

cases:
– when the dictionary explicitly mentions the word as being coined

in SF literature. Such is the case with terraforming, created by
Jack Williamson in Collision Ship (1942), which merits entries in
both Wiktionary and Wikipedia, as well as with conapt, coined
by Philip K. Dick and taken up by other SF authors. We also
kept the word psionic (as well as its French equivalent psionique )
because the French Wiktionary entry explicitly links it to SF
literature.
– and, finally, when the meaning of the word was clearly neological,
such as: optimizer (Barnes Mother of Storms, 1996) which there refers
to a kind of AI.
Because some works (for instance those by Asimov or Ayerdhal) proved

to be much richer in fiction words than others, the number of different
fiction words per author was capped at a maximum of 15 to prevent a
possible bias favouring an individual author’s style in our analysis. (Note
that several occurrences of the same fiction word represent only one
type.) When the same fiction word appeared in the works of two differ-
ent authors, we kept both types on the list, because it seemed interesting
to identify neologisms used by multiple authors.
When filtering ended, the lists included 120 and 133 French and
English fiction words, respectively. To obtain comparable figures, we
randomly dropped 13 English fiction words. The two final lists there-
fore contained 120 words each, coined by 15 different authors in
French and 19 different authors in English.
158
The methodology described above cannot tag compounds sepa-

rated by spaces or dashes, <FW> lemmas being identified at the token
level. Compounds were thus ignored to ensure focusing exclusively
on lexical creations without separators only. It can be assumed, how-
ever, that the following observations would also apply to multi-word
compounds.
3.2 Preliminary Comments
Before discussing semantic and morphological aspects of fiction words,

we will discuss another notable phenomenon: the sharing among
authors of a significant set of fiction words. The existence of a kind of a
common SF terminology is well-known, and, in fact, as Prucher (2017)
notes,7 even many words and scientific terms in general use today were
originally coined in SF literature (robotics, genetic engineering, zero-g,
deep space, ion drive, pressure suit, computer virus and worms, gas giant, et
cetera).
The following examples are from the lists of extracted words:
Astroport sees widespread use by at least 10 different French authors
at last count: Ayerdhal, Brussolo, Dantec, Deff, Dunyach, Genefort,
Lehman, Leourier, Ligny, and Wagner. The term is absent from both
our reference dictionaries and the IATE terminology database but is
listed in the GDT,8 which could explain its wide dissemination. The
SF authors clearly prefer the term to its official variant spatioport (34
vs. 159 occurrences), which is the one the dictionaries reference. It
may not be a fiction word (one would have to know its first attes-
tation to clarify its status), but it is notwithstanding a popular term
in fiction, unlike its non-fictional competitors. The English equiva-
lent, spaceport, is well-referenced in dictionaries and, just like astroport
7See the post on https://io9.gizmodo.com/5850293/10-words-you-might-think-came-from-sci-
ence-but-are-really-from-science-fiction.
8See http://www.granddictionnaire.com/ficheOqlf.aspx?Id_Fiche=8349051.
159
in the French corpus, is quite common in the English corpus (119

occurrences).
Astrogator is used by three authors in French and one author in
English (in its astrogation variant).
Conapt, borrowed from Philip K. Dick, who coined the word, is used
by two French authors (Ligny and Genefort).
Holo, as either noun or adjective is used by 6 authors in French and
11 authors in English (a holo scene, a holo simulation ) to designate a hol-
ogram by means of an apocope. This wide diffusion should be compared
with the very large number of fiction words with holo as their root (with
its specific meaning of hologram, which is not the meaning of the Greek
root holo in classical compounds):
English: holomovies, holosign, holovision, holotapes, holoterminal,

holovid
French: holosite, holopub, holoporno, holorama
Holo is one of the most productive roots on our extract lists and appears
to be a distinctive marker of the SF genre.
Generally speaking, the dissemination of these fiction words shows
that the authors in question are inclined to situate their novels in a
common fiction space shaped by the major references within the genre
tradition. It should also be noted that some authors in the present cor-
pus of novels—Brin, Butler, Pohl, and Robinson for English or Pagel,
Di Rollo, and Fontana for French—have coined very few fiction words.
A book’s theme is not always the key to explaining a dearth of fiction
words: some novels are rich in technical or scientific innovations, but
the authors prefer uncommon specialized terms of scientific origin:
for example, Brin (Earth, 1990) who uses terms such as stirpiculture,
hypergolic, supercold, spacesickness, and cryocanister. It is a well-known
fact that scientific and technical terms abound in SF literature—some
authors even suggest using these novels as an interesting resource for
developing scientific literacy in L2 learners (Rolls and Rodgers 2017).
Indeed, these scientific terms in science fiction often play the same role
160
as fiction words, since they are unfamiliar to readers and thus tend to
produce in them the same sense of cognitive estrangement. We will
take this point up again below with examples from the corpus (see
Sects. 4.2.1 and 4.3).
3.3 Semantic Annotation
Making this a corpus-driven study involved the choice of not starting

from an a priori categorization: the selected fiction words were instead
grouped according to semantic affinities and then assigned, each time
a group exceeded 5 elements, to one of the distinct categories shown in
Table 29:
This categorization procedure was intuitive and interpretative and
could probably be refined. Especially TECH, the richest category, could
be further divided into subcategories: material, equipment, vehicles,
communication tools, computing, robotics, et cetera. And, of course,
the individual categories have fuzzy boundaries: for instance TECH
and SCI at times may be nearly indistinguishable, just as the distinction
between POP and FUNC may be anything but obvious (datavandal
could belong to both categories, depending on the context). Still, even
a rough typology has the advantage of revealing the strong tendencies
more clearly. Figures 1 and 2 show the distribution of fiction words for
the French and English corpora:
The similarities between these distributions are striking, particularly
for the two main categories TECH and FUNC. POP is also one of the
main categories, which is not surprising given its proximity to FUNC
(both categories refer to groups of individuals). For the other catego-
ries, the numbers are probably too low to allow drawing general con-
clusions: for example, for English, half of the fiction words in the ANI
category come from just one animal fiction work—Crichton’s Jurassic
Park (1990).
9For an example of applying a complex semantic grid to a science fiction corpus, see Chapter 7 by
Goossens et al. in this volume.

Table 2 Semantic classes derived from the selected fiction words
Category Glossary Examples in English Examples in French
6
LOC Name of public or private place decel, helldeck, cemeterium astroport, arcologie, conapt,
comsal
TECH Vehicle, material, software and bublepak, ornithopter, passtouch, agrave, plastacier, holorama, voc-
digital concepts, technological nervoplex, holosign, biomech, odeur, synthépoutres, téléporta-
product, robot, communication holotapes, replicator trice, neurofibres, holoporno
device
FUNC Profession, practice, function gengineer, nanoarchitect, astrogateur, cyberurgie, ordino,
noncoms, datavandal, décyb, vidéovamp
psychohistorian
POP Name of people, race, species, lan- empath, posthuman, jagernaut, galla, érudes, transvers, gurde,
guage, community, group hsai thoréide
BIO Biological particularity, disease, metapheromon, metapheromonal sexomorphe, térato-frères,
mutation, biological product épidermie
PSY Psychic faculty, psychic syndrome psionic, precog, hyperempathy, psionique, métanoia
mentalic
SUB Edible substance, beverage, food, hyperdex, trank, soltoxin, chimeïscine, amphécafé,
drugs, poison nutriphore nutripoule, glucogel
GEO Derived from a place name loumkane, scorpiique, méladorien,
marsilien
ANI Animal postcanine, scions, ecos, procomp- buccins, cyanosaure, chélide,
sit, othnielian ptéroxanthe
VEG Plant dayvine gleis, lagad, janéhilia, véism,
lépidodendron
SCI Scientific name for natural microstring, subetheric, realspace champ-neg, introns, exons
phenomena
POL Political function, political party, telesenator, cooption interstellariste, ucdu, technotrans,
fap
French and American Science Fiction During the Nineties …
government organization,
161
corporation
162
'LVWULEXWLRQRIILFWLRQZRUGVE\VHPDQWLFILHOG)UHQFKFRUSXV
Kd,͕ϱ ^/͕ϯ
W^z͕ϯ Kd,
'K͕ϰ
^/
WK>͕ϰ
W^z
^h͕ϰ 'K
d,͕ϰϲ
WK>
/K͕ϰ
^h
E/͕ϰ
/K
E/
s'͕ϲ
s'
>K
>K͕ϭϭ WKW
&hE
d,
&hE͕ϭϯ WKW͕ϭϯ
Fig. 1 Distribution of the selected words in the French corpus (n.b. the OTH
label refers to words that have not been categorized)
3.4 Morphological Annotation
The distribution of lexical categories once again presents very similar pro-
files to the observer: overall, nouns clearly dominate, accounting for more
than 82% of the fiction words, followed by adjectives and adjectives used as
nouns corresponding to the FUNC and POP categories (posthuman, men-
talic, hsai for English, or yrvène, sexomorph, gynoïde for French) (Table 3).
Concerning the various modes of word formation, we annotate the
following cases:
– PCL: purely classical compounding (using only Greek and Latin

roots)
EN: ornithopter, subetheric, nutriphore
FR: sexomorphe, transvers, lepidodendron
163
'LVWULEXWLRQRIILFWLRQZRUGVE\VHPDQWLFILHOG(QJOLVKFRUSXV
Kd,͕ϲ 'K͕Ϭ
s'͕ϭ WK>͕Ϯ
Kd,
/K͕ϯ
>K͕ϰ 'K
^h͕ϰ s'
^/͕ϰ WK>
/K
W^z͕ϲ >K
d,͕ϱϳ
^h
WKW͕ϳ ^/
W^z
WKW
E/͕ϭϬ E/
&hE
d,
&hE͕ϭϲ
Fig. 2 Distribution of selected words in the English corpus
Table 3 Comparative distribution according to POS

POS EN FR
Nouns 99 98
Adjectives used as nouns 9 10
Adjectives 8 11
Verbs 2 1
Adverbs 1 0
– HCL: hybrid classical compounding with a Greek or Latin root and

an English or French root
EN: posthuman, nervoplex, cyberlock

FR: chromobéton, neuroséquenceur
164
– COMP: popular compounding with modern roots
EN: flycar, passtouch, bublepak, datarodent

FR: crédibloc
– PM: portmanteau words consisting of the fusion of two truncated words
EN: fibrop (optical + fibre), graviscan (gravitational + scanner), tricom-

puter (3D computer), gravicab (gravitational cab)
FR: plasticuir (plastique + cuir), vocodeur (codeur + vocal), plastacier
(plastique + acier), holopub (publicité + holographique)
– PM2: opaque portmanteau words, formed in the style of a brand

name
EN: soltoxin, cavitron
– DER: derived words:
EN: experiencer, optimizer

FR: téléportatrice, exploreur, hégémoniaque
– ABR: abbreviations
EN: grav
FR: champ-neg, sim (for simulation), holo
– INI: acronyms
FR: ucdu, fap
– PARO: paronymic distortions of attested words
EN: jagernaut (for juggernaut)

FR: dollus (for dollard), trax (tractor)
165
– BOR: borrowings
FR: sample (in the sense of avatar), outzone, outer
– NEO: lexical creations (to which some suffixes may be attached)
EN: trank, hsai

FR: yrvène, gleis, véism, sukmen
Figures 3 and 4 show the distribution of these types of formations in

the French and English corpora.
In this respect, the English and the French corpora profiles differ sig-
nificantly: classical compounding, well-attested to in the French corpus
(28 fiction words) plays only a minor role in the English corpus. On the
'LVWULEXWLRQRIW\SHVRIIRUPDWLRQ )UHQFKFRUSXV
KDW͕ϭ WZK͕Ϯ
WDϮ͕Ϭ
KZ͕ϯ
Z͕ϳ
W>͕Ϯϲ KDW
Z͕ϵ WDϮ
WZK
KZ
/E/͕Ϯ Z
Z
WD
WD͕Ϯϳ EK
,>͕Ϯϭ ,>
/E/
W>
EK͕Ϯϭ
Fig. 3 Distribution of types of formation for the selected words in the French
corpus
166
'LVWULEXWLRQRIW\SHVRIIRUPDWLRQ (QJOLVKFRUSXV
WZK͕Ϯ
KZ͕ϭ
WDϮ͕ϳ
KDW͕ϯϬ W>͕ϴ KZ

WZK
WDϮ
Z͕ϭϬ
W>
Z
Z
Z͕ϵ
EK
WD
EK͕ϲ ,>
,>͕Ϯϯ
KDW
WD͕Ϯϭ
Fig. 4 Distribution of types of formation for the selected words in the English
corpus
other hand, popular compounding, predominant in the English corpus,

is mostly lacking in the French corpus. This may be a consequence of
the overall structure of the scientific-technical vocabulary, which con-
tains popular compounds more often in English than in French, par-
ticularly for digital technology and innovation (with words such as
powerchip, mainframes ), but this hypothesis needs further investigation.
Moreover, our decision to drop compounds with dashes or spaces from
this study could explain the absence of popular compounds for French.
However, in one respect the two corpora exhibit a striking similar-
ity: truly opaque neologisms, such as hsiel (EN) or gleis (FR), are very
rare. In most instances, derivational morphemes (FR yrvène, marsilien )
or flectional morphemes (callies ) provide clues for decoding the word
even for the NEO categories. The strangeness effect of the xenoency-
clopedic world is to be found more in referents and concepts than in
morphemes.
167
Another shared and rather striking practice involves the use of abbre-
viations. We encountered this phenomenon in both the PM and the
ABR categories, which total 30 and 34 fiction words in the English and
French corpora, respectively. The tendency to shorten words leads par-
ticularly to creating new abbreviated stems related to their classical or
popular sources and placed in the prefix or suffix position. Heavily rep-
resented stems are nano- (with the meaning of nanotechnology), psy-,
grav-, med-, cyb-, com-, holo-, info-, or -tron, -vid, -cog, -tech, -cyb, and
-net. This may have two possible causes: on the one hand, shortening
certain words, such as holo (for hologram ) or grav (for gravity ), produces
a familiarity effect; on the other hand, it may indicate a dialectical game
of juxtaposing the banal with the extraordinary. The inversion of these
poles and the loss of reference points is one of the SF genre’s character-
istics. As observed by Landragin (2018, 138), “l’abréviation sert claire-
ment à nous montrer que le mot est tellement courant dans la langue
du monde fictif qu’on l’emploie sous une forme abrégée, comme nous
le faisons tous pour parler du métro, d’un resto ou d’un ciné” (abbre-
viation is clearly used to show us that the word is so common in the
language of the fictional world that it is used in an abbreviated form, as
we all do when talking about the métro [instead of métropolitain ], a resto
[instead of restaurant ] or a ciné [instead of cinéma ]; our translation). In
addition, it should be noted that this type of coinage is widespread in
the context of industrial innovation, especially for brand naming, which
makes extensive use of the truncated stems mentioned above. Indeed,
several fiction words on our lists are attested to as names of products,
organizations, companies, or laboratories (genemods, medtech, scalpnet,
bublepak, cavitron, etc.).
4 The Contexts of <FW> as a Pivot:

Analyses of Four Prepositional Phrases
Most of the extracted RLTs correspond to colligations, that is, co-oc-
currences of the <FW> units with grammatical units such as determin-
ers, prepositions, auxiliary verbs, conjunctions, et cetera. This was to be
168
expected, because the <FW> class is very large and does not really corre-
spond to a semantically consistent paradigm.
As a preliminary observation, <FW> pivots appear to be frequently
associated with coordination, in both the English and the French corpora:
<<FW> and <FW>> occur 86 times in English:
(6) … only then did they remember that they had not brought their
working tools, their handcomps and shirtcoms, to a quiet dinner in the
small house. (Moon Remnant Population, 1996)
<<FW> le <FW>> occurs 17 times in French:
(7) Inutile de sonder le svøn ou l’urgal – tous deux s’avèrent d’une grande
pauvreté lorsqu’il s’agit d’exprimer aussi bien des sensations que des con-
cepts abstraits. (Wagner Le Chant du cosmos, 1999) (No need to probe the
svøn or the urgal – both are very poor when it comes to expressing both
sensations and abstract concepts.)
Coordination may have different functions: subsuming the various

fiction words under the same hypernym, as in (6); sharing the same
explanatory context, as in (7); or reinforcing the effect of a sense of
wonder by means of an accumulation of exotic details, as in (8):
(8) Les mercenaires de Pritiv poussèrent sans ménagement les danseurs

de mazakawen, les agres, les féelles et les deux majikens, les officiants de
magie, vers les coulisses. (Bordage Les Guerriers du silence, 1993) (The
Pritiv mercenaries pushed the mazakawen dancers, the agres, the féelles
and the two majikens, the magic officers, unceremoniously backstage.)
Another interesting finding is the recurrence of predicative structures in

French, with 200 occurrences of <le <FW> être>:
(9) La samsonite n’est ni contagieuse, ni mortelle. (Ligny Albatroys, 1991)

(Samsonite is neither contagious nor deadly.)
(10) Les sigres sont inoffensives. (Genefort Arago, 1995) (The sigres are
harmless.)
169
Most of these occurrences are encountered in dialogues or reported dis-

course, which would seem to be ideal for introducing explicitly “didac-
tic” elements (see Sect. 4.2). This is confirmed by the frequency of the
RLT <que le <FW> être>, which appears 34 times, in completive subor-
dinates such as:
(11) On dit que les kroaz sont les pires des êtres vivants. (Bordage Les
Fables de l’Humpur, 1995) (It is said that kroaz are the worst of living
beings.)
By contrast, in the English corpus <the <FW>> appears to be negatively

correlated with the verb to be: the association score (loglike) is about
−75, which indicates that the number of actual co-occurrences is far
lower than one would expect. Is that kind of didactic segment regarded
as too explicit in the English language? Is it a matter of stylistic prefer-
ence? We defer to future investigations to provide a clear interpretation
for this as-yet unexplained contrast.
After these preliminary observations, the main finding is that for
both English and French the clearest colligation statistics single out
prepositional phrases. We will elaborate on these observations in the fol-
lowing section.
4.1 Description of the Contexts for Appearances

of <FW> Pivots
4.1.1 Methodology
From the list of <FW> pivot relationships, we chose the most frequently
found one (and the one that seems most likely for linguistic reasons) for
our analysis: the <FW> noun used with a definite article (le, la, les and
the ). From this very large group of units (6779 for EN and 3995 for
FR), prepositional phrases formed according to the pattern <PREP_the/
le_<FW>> in turn were selected for a more detailed analysis. This meant
making a statistically based decision: the most representative preposi-
tional phrases were chosen for study, that is, <of the <FW>> (812 times,
170
dispersion 48) and <in the <FW>> (396 times, dispersion 46) for the
English corpus; <par le < FW>> (by the <FW>, 112 times, dispersion
30) and <sur le <FW>> (on the <FW>, 114 times, dispersion 29) in
French. These numbers had to be reduced for the following reasons:
some <FW> might have been incorrectly tagged because of noisy tokeni-
sation (for instance “bricà” for bric-à-brac, “ing” for smil-ing ). Some
words were not on our reference lists despite their existing in French or
English albeit as seldom-used or outdated derivatives. Thus, we culled
a number of unusual words from our lists (par les craillements, sur les
putasseries ), including some expressions from specialized lexicons: sur les
espars (nautical word), par le catgut (surgical suture), the different kinds of
smectites (geological term), or, for English, compound names or borrow-
ings such as of the amphitheater, of the demi-mondaine. For the French
corpus, noise constitutes about 10% of the occurrences of the two prep-
ositional phrases studied and approximately 25% in case of the English
corpus.
Based on these four lists of occurrences, we chose to keep 50 for
each prepositional phrase. We aimed to vary the instances of <FW> in
a prepositional phrase in terms of both lexical morphology and refer-
ent: names of peoples, plants and not just technology or ships, those
defining themes closely affiliated with the SF subgenre in the popular
imagination.
4.1.2 Topics Represented by the Nouns <FW>

in the Four Prepositional Phrases
For the three prepositional phrases <par le < FW>>, <sur le < FW>>,
and <of the < FW>> it was easy to select fiction words referring to spe-
cies of living beings (POP, ANI and VEG categories: names of “races”;
FUNC category: social roles): par les deks (Bordage Abzalon, 1998), of
the phytids (Bear Legacy, 1995). Fiction words may also refer to diseases
or drugs (BIO and SUB categories): sur l’épidermie (Berthelot Rivage
des intouchables, 1990), of the hyperdex (Silverberg Hot Sky at Midnight,
1994). The prepositional phrase <in the < FW>> dominates the others
in the representation of transportation (mainly ships: TECH category),
171
and obviously of places (LOC category: buildings, rooms): in the drop-

ship (Simmons The Fall of Hyperion, 1990), in the flawship (Bear Legacy,
1996).
Each of the four prepositional phrases selected include words coined
as references to power systems or their representatives (POL category):
par les Éliges (Ayerdhal L’Histrion, 1993), in the areophany (Robinson
Blue Mars, 1996). In the English corpus, <FW> nouns designating vir-
tual products are more likely to occur (TECH category: software, sim-
ulations, etc.): of the ractives (Stephenson Snow Crash, 1992), in the
stimsim (Simmons Endymion, 1996).
4.1.3 The Most Frequent Syntactic Functions
Among the 50 selected examples, the most frequent syntactic function

for <par le <FW>> is that of an agent phrase. It follows either a conju-
gated passive form or a past participle used as an adjective. Thus, the
action is carried out <par le <FW>>:
(12) Elle avait failli être emportée par l’estérionite un an plus tôt, et
seul l’amour d’un jeune homme du nom d’Arel, l’arrière-petit-fils d’un
dek qui avait bien connu Abzalon, l’avait raccrochée à la vie. (Bordage
Abzalon, 1998) (She had almost been carried off by the esterionite a year
earlier, and only the love of a young man named Arel, the great-grandson
of a dek who had known Abzalon well, had kept her alive.)
The estérionite is probably an epidemic, a disease: être emporté par and

the suffix -ite allow this inference (in French, the suffix -ite is used to
form many terms in the field of medicine: appendicite, bronchite ). The
stem of the derived word is the proper noun Ester, which refers to
the fictional world in which this novel is set. The reader can therefore
proceed on the basis of the principle of what Langlet (2006, 27) calls
“abduction”: unlike deduction or induction, “abduction” here means
formulating a hypothetical rule, which will be either confirmed or inval-
idated in the course of the reading process. Of course, the sentence
172
quoted above is not necessarily the first appearance of this fiction word
in the novel.
For this prepositional phrase, few expanded noun phrases exist in the
French corpus: the <FW> noun often stands alone, and this form seems
generally more suitable for narration than description. The analysis of
the 50 selected occurrences leads to the conclusion that this preposi-
tional phrase provides an idea of the type of actions usually performed
by unknown characters or of objects represented by a fiction word.
The latter is typically described by its effects. This syntactic function is
a dynamic component of the description, essentially oriented towards
standard actions:
(13) Outre son rôle de superviseur des enquêtes effectuées par les loyals
de sa cellule, un justicier avait également pour mission de veiller à leur
parfaite légalité, ainsi qu’au respect des droits imprescriptibles de tout
être humain ou animal impliqué. (Wagner Le Chant du Cosmos, 1999)
(In addition to his role as supervisor of investigations conducted by the
loyals of his cell, a justiciar was also responsible for ensuring that they were
perfectly legal and that the imprescriptible rights of any human or animal
being involved were respected.)
In this example, the prepositional phrase par les loyals is an agent phrase
of the past participle effectuées: the noun enquêtes indicates the stand-
ard activity of the loyals, and the complement de sa cellule specifies the
organization of the investigators into units headed by a supervisor called
a justicier. Thus, the agent phrase allows us to define a category of char-
acters in action.
The most frequent syntactic function of <sur le <FW>> is that of
indirect object:
(14) Virus entrouvre enfin ses paupières boursouflées, pose sur le decyb
un regard empreint d’un soupçon de lucidité. (Ligny Cyberkiller, 1993)
(Virus finally opens her swollen eyelids, glancing at the decyb with a hint
of lucidity.)
Several prepositional phrases <sur le <FW>> also are used as noun

complementation:
173
(15) Pendrek contacta Plaike. Il lui parla d’un rêve qu’il avait fait à la suite
de son discours sur les fulguriers. (Genefort Arago, 1993) (Pendrek con-
tacted Plaike. He told him about a dream he had following his speech on
the fulguriers.)
The fiction word when introduced by the preposition sur has more com-
plementation than when introduced with par. These complements are
in specific relative clauses and epithets, which make it possible to char-
acterize the unknown noun. This is particularly true when it comes to
descriptions:
(16) Ils s’extasièrent bruyamment … sur les albotoès aux frondaisons mul-
ticolores, … et enfin les rarissimes arborivoles dont les cimes flottantes et
reliées au sol par de fines et souples lianes transparentes surlignaient de
mauve cette fabuleuse luxuriance végétale. (Bordage Les Guerriers du
silence, 1993) (They were loudly ecstatic … about the albotoès with their
multicoloured foliage, … and finally … the rare arborivoles whose float-
ing tops linked to the ground by thin and supple transparent vines, high-
lighted with mauve this fabulous plant luxuriance.)
The fiction word albatoès has a complement that lets us deduce that it is
a tree: this complement is frondaisons (foliage). The latter is part of the
albotoès, and receives a qualification: multicolores (multicoloured). This
is therefore an aspectualization operation (Adam 2017, 94–97), which
consists of a descriptive sequence of fragmenting the whole into parts
that are characterized. Similarly, arborivoles—another fiction word—is
qualified by an adjective indicating a high degree—rarissime (rare)—
which reinforces the unknown nature of the name, itself characterized
by a relative clause detailing its parts (cimes, lianes, mauve ) in another
aspectualization operation.
However, the prepositional phrase <sur le <FW>> is no less narrative
than <par le <FW>>: when it functions as an indirect object, it is most
often the object of violence: braquer (aim), tirer (shoot), s’acharner sur
(rage over)…; the goal of a movement: se précipiter sur (rush on); or the
object of a look: les yeux se posent sur (eyes are set on), poser un regard sur
(take a look at).
174
The most frequent syntactic function of <of the <FW>> is the com-
plementation of quantifiers: quantity + <of the <FW>>, indicating that
a certain number is singled out from a whole, which is represented
by <FW>:
(17) Edible after soaking in water and cooking, high in usable protein
and sugars, sweet and meaty to the taste, diospuros had been one of the
first phytids used successfully for food. (Bear Legacy, 1995)
The expression one of the first phytids used successfully for food indicates a
part of the entire group of phytids. The fiction word has complementa-
tion: the adjective first and the participial phrase specify its characteris-
tics. The noun phrase is the complement of diospuros, another fiction
word, and a hyponym of phytids. To a lesser extent, the prepositional
phrase <of the <FW>> functions as a noun complementation.
In the phrase <of the <FW>>, the fiction word has a specific comple-
mentation, especially with ordinal adjectives (earliest, first ) or adjec-
tives relating the individual to a group (most common in the example
below):
(18) The Skroderiders were one of the most common sophonts in the Beyond.
There were many varieties, but analysis agreed with legend: very long ago
they had been one species. (Vinge A Fire Upon the Deep, 1992)
Here, we can observe the use of the noun phrase as a subject comple-
ment, the same as in example (17): the fiction word in the preposi-
tional phrase <of the <FW>> is a hypernym of the subject. Fiction words
thus make it possible to define one another in an inclusive link, either
because an individual or a group of individuals are part of a category
or because they are not. These observations confirm the importance of
the POP and FUNC fiction word categories (Sect. 3.3): science fiction
often portrays an individual who represents his origins and his function
and then liberates or tries to liberate himself from both.
The most frequent syntactic function of the prepositional phrase <in
the <FW>> is that of a place adjunct:
175
(19) Ry and I put on light pressure suits. He picked up a box not much
larger than his head and tucked it under his arm. We nodded to an eye
conveying our images to the pilot waiting in the flawship above; then we
stepped outside. (Bear Legacy, 1996)
In several occurrences, the prepositional phrase is also a complement of

the noun whose location it indicates:
(20) If the psions in the powerlink were too similar, it set up a resonance
like a driven oscillator, forcing their minds into greater and greater fluctu-
ations until the link shattered. (Asaro The Veiled Web, 1999)
With this specific prepositional phrase, there is usually little comple-

mentation of the fiction word, which often appears alone. It is also
more difficult to find fiction words referring to species using this con-
struction: the fiction words mainly designate rooms, buildings, or
modes of transportation, especially spaceships, some of which carry
entire populations: freezatorium (Pohl The World at the End of Time,
1990), seedship (Simmons Endymion, 1996).
The selected prepositional phrases, which are statistically representa-
tive of the use of a fiction word with a definite article, only rarely repre-
sent the first appearance of the unknown word in a novel. Still, some of
the 200 occurrences we analysed were framed by explanatory contexts
that are interesting to look at in more detail.
4.2 Didactic Segments
The prepositional phrases selected for further analysis are sometimes

inserted in explanatory contexts, which make it possible to under-
stand the fiction word and determine its referent. The relevant passages
in the narratives are “didactic segments”: coming in varying lengths,
they accompany the fiction word to give a definition or at least “une
clé de compréhension” (Langlet 2006, 27) (a key to understanding; our
translation).
176
4.2.1 Appositions
The most straightforward of these didactic mechanisms is the apposition

(Langlet 2006, 38–39): it is the simplest grammatical tool for inciden-
tally defining a fiction word without disturbing the narrative flow.
(21) Dans le lointain se profilait un océan de dunes, vagues blafardes et

pétrifiées surmontées d’une écume de sable soulevée par le chounza, un
vent sec et froid descendu des montagnes du grand Erg Brûlé. (Bordage Les
Guerriers du silence, 1993) (In the distance lay an ocean of dunes, pale
and petrified waves surmounted by a foam of sand raised by the chounza,
a dry and cold wind coming down from the mountains of the Great Burnt
Erg.)
The apposition un vent… assigns the fiction word chounza to a cate-

gory (to a hypernym). The apposition thus allows the reader to fall back
upon his general encyclopaedia to feed his xenoencyclopedia, thanks to
an analogy suggested by the text.
Appositions turn out to be such handy devices that some authors
even use them in series:
(22) Des chercheurs d’optalium, un métal rare très prisé par les sculp-
teurs-joailliers de Bella et les corporations de l’artisanat sacré de Marquinat.
Des types rongés par la zénoïba, la fièvre des temps de pluie, une maladie
incurable. (Bordage, ibid. ) (Searchers for optalium, a rare metal highly
prized by the sculptors-jewellers of Bella and the guilds of the sacred crafts
of Marquinat. Guys tormented by zénoïba, rainy season fever, an incurable
disease.)
The first apposition defines optalium on the basis of the use of hyper-
nymy: it is a type of metal. One can see the cliché of rarity—rare—
reappearing (see in the same novel the use of rarissimes, example 16).
The second apposition allows the reader to define zénoïba, for which the
syntactic context makes an abduction possible even before reading the
apposition: the agent phrase referring to individuals after a past partici-
ple suggests that the unknown name is either a passion, or a disease, a
177
virus. The apposition then specifies the disease in question: la fièvre des
temps de pluie (rainy season fever), only to be followed by another appo-
sition, une maladie incurable (an incurable disease). At this point, the
noun maladie (disease) is superfluous, but the adjective incurable adds a
trait to the intensity of zénoïba.
The following case is particularly interesting in the use it makes of the
determiner introducing the apposition:
(23) Et, d’ailleurs, que s’était-il passé à l’époque? Tem lui-même n’en
avait qu’une idée très vague. Il disait que, pour des raisons inconnues, la
Psychosphère avait commencé à déborder sur la Réalité consensuelle. Ce
phénomène était peut-être dû à des perturbations causées par le semen
of gods, cette drogue employée autrefois pour voyager télépathiquement.
(Wagner Le Chant du Cosmos, 1999) (And, by the way, what had hap-
pened at the time? Tem himself had only a very vague idea of it. He said
that, for unknown reasons, the Psychosphere had begun to overflow into
Consensual Reality. This may have been due to disruptions caused by the
semen of gods, this drug formerly used for traveling telepathically.)
The apposition is a noun phrase introduced by the demonstrative deter-

miner cette (this); the noun drogue has complements. This expanded
noun phrase is analogous to the form “ce… qui… + present tense”
studied by Éric Bordas, who sees in it a realism marker (for example
in Balzac’s novels). The construction above is a variant: the demonstra-
tive determiner is a “démonstratif de connivence” (Bordas 2001, 33)
(demonstrative of complicity; our translation). It refers to a memory
supposedly shared by the narrator and the reader (in a memorial exo-
phora), bringing the latter into the universe of the story. Thus, the nar-
rator presents information about the SF world while suggesting that this
knowledge has, in fact, already been established by the reader.
All examples given in this section were taken from the French corpus;
as it happens, the English corpus exhibits fewer appositions for defin-
ing fiction words, at least among the 200 occurrences that we retained.
There are, however, some appositions in the corpus which define words
borrowed from other languages:
178
(24) Of all the members of their cabal, she had been the first struck per-
sonally by the lashing tail of the taniwha – the monster in the Earth’s core.
(Brin Earth, 1990)
(25) The stratiform copper deposit that they had been raking up ran dry,
and it was time for another ráhla, the movement of the hejra to the next site.
(Robinson Red Mars, 1992)
The taniwha is a creature borrowed from Māori mythology, and the

ráhla is a movement of nomadic Arab tribes. However, these words can
easily be (mis)interpreted as fiction words by a reader who has never
heard them before. Samplings conducted in the English and French
corpora showed that authors of SF novels coin fiction words but also
borrow existing words from fields of expertise that are sometimes so
specialized that the reader is likely to categorize them as fiction words
(as mentioned in Sect. 3.2): cases in point include (in the English cor-
pus) cyanophyte (type of bacteria), smectite (type of mineral), rectenna
(for rectifying antenna ) and (in the French corpus) tores (type of mould-
ing in architecture), carabes bombardiers (type of beetle). These words
already have meanings in the readers’ world although the average reader
is likely to draw a blank on encountering them. SF authors use apposi-
tions to make these words accessible in their narrative, and they do the
same with fiction words, some of which are formed through precisely
the same morphological mechanisms as specialized words (PLC and
HCL formations, see Sect. 3.4). Thus, one may mistake a fiction word
for a specialized word and vice versa: essentially, the same phenomenon
occurs when a demonstrative determiner introduces the apposition.
4.2.2 Explanatory Context
Fiction words can also be explained by using a segment that is longer

than an apposition. Here are some examples of explanatory contexts
suitable for embedding fiction words.
Firstly, heterodiegetic narrators sometimes intervene directly, inter-
rupting the action with an explanatory remark:
179
(26) Là réside aussi la supériorité de Deckard sur les autres decybs: il n’a pas
oublié l’ancienne culture, les façons de vivre de jadis, et en tire toujours
un enseignement. (Ligny Cyberkiller, 1993) (This is also where Deckard’s
superiority over the other decybs lies: he has not forgotten the old culture,
the ways of life of the past, and he always learns from it.)
The prepositional phrase sur les autres decybs (over the other decybs)
makes it possible to individualize the hero in relation to his peers (les
autres decybs, plural). It stresses his supériorité to show his uniqueness.
By explaining the quality that sets Deckard apart, the narrator simul-
taneously defines a whole category of characters in negative terms.
Narrative heterogeneity is quite strong here as far as the contrast
between the plot level on the one hand and the commentary of an
extradiegetic and omniscient narrator on the other is concerned.
Secondly, the explanatory statement can also be attributed to one
of the characters, whether or not this character is the homodiegetic
narrator:
(27) Something had changed in the divaricates on their arrival in

Lamarckia. Lenk had encouraged new births, of course. But divaricates
had generally had no more children on Thistledown than other Naderites,
no more even than most Geshels. On Lamarckia, having children had
become a ruling passion, as if some hidden drive had been awakened,
and the human race – isolated as this weak little seed on a huge world –
had needed to spread its limbs and foliage far and wide once more. (Bear
Legacy, 1995)
The section before this extract consists of Shirla’s reported speech. The
narrator’s thoughts draw conclusions from this reported discourse.
The explanatory context is based on general laws that are infringed on
Lamarckia. And, the narrator proposes an analogy that makes it easier to
understand the demographic phenomenon. In this case, the inner speech
takes over from a character’s speeches to extend the explanatory segment.
Thirdly, when the story is told in the first person by the protagonist,
the explanatory speech probably fits more logically into the narrative:
it embraces the hero’s questioning. The occurrences of <PREP_definite
180
article_ <FW>> from the English corpus provide two examples of meta-
linguistic segments. In the first, the autodiegetic narrator develops a the-
ory about the difficulty of describing unknown living organisms (before
eventually describing them):
(28) Before I attempt to describe the things, I have to say that little in the
history of humankind’s expansion in this arm of the galaxy had prepared us
to describe large alien organisms. On the hundreds of worlds explored and
colonized during and after the Hegira, most of the indigenous life discov-
ered had been plants and a few very simple organisms, such as the radiant
gossamers on Hyperion. The few large, evolved animal forms – the Lantern
Mouths on Mare Infinitus, say, or the zeplins of Whirl – tended to be
hunted to extinction. The more common result was a world filled with a few
indigenous life-forms and a myriad of human-adapted species. Humanity
had terraformed all these worlds, bringing its bacteria and earthworms and
fish and birds and land animals in raw DNA form, defrosting embryos in
the early seedships, building birthing factories in the later expansions. The
result had been much as on Hyperion – vital indigenous plants such as the
tesla trees and chauna and weirwood and some surviving local insects coex-
isting with thriving Old Earth transplants and biotailored adapts such as tri-
aspen, everblues, oak trees, mallards, sharks, hummingbirds, and deer. We
were not used to alien animals. (Simmons Rise of Endymion, 1997)
The extensive explanatory segment includes the history of living organ-

isms since “the history of humankind’s expansion in this arm of galaxy”.
The prepositional phrase in the early seedships occurs in this segment,
putting into perspective the history of galactic colonization, itself rich
in fiction words; the historization appears in the adjective early, which
places this episode at the beginning of the terraforming. In this exam-
ple, only the beginning is metalinguistic: it is followed by an explan-
atory account of the origins. But in the other example, the whole
didactic segment is metalinguistic. The fiction word is then treated first
as an autonym whose definition the narrator provides:
(29) I labored to memorize – and to forget some of what I had learned on

the Lake of the Winds. / Fortunately, on the upper decks and belowdecks,
the names had changed little. Fore and aft still applied: bow, midships,
181
stern; forecastle, foredeck or maindeck, quarter deck aft of the maintree,

but the poop aft of the mizzen had reverted to the original Latin, pup-
pis. The long superstructure on the puppis, which appeared top-heavy to
me, was called, with affection, the pupcastle. On the Vigilant, the captain,
master, doctor, and researchers kept quarters here, and the two laborato-
ries were also in the pupcastle. (Bear Legacy, 1995)
The final prepositional phrase is not the first time the fiction word is
mentioned: we must go up a few lines to find the autonymous use,
The long superstructure … was called … the pupcastle. The homodiegetic
narrator comments here on the nautical terms on the ship he has just
joined. The fiction word is therefore defined by a narrator who is not
familiar with the terminology used on board either and is discovering
it at the same time as the reader. More often, however, the narrator
already has knowledge about the SF world and delivers it to the reader
as soon as strangeness threatens understanding:
(30) An archangel-class starship translated into God’s Grove space the day
after Pope Julius’s death. … / Two men and a woman were aboard. Their
presence in the dropship was a curiosity – the archangel-class starships
invariably killed human beings during their violent translation through
Planck space and the onboard resurrection crèches usually took three days
to revive the human crew. / These three were not human. (Simmons Rise
of Endymion, 1997)
The narrator here is able to introduce the presence of the three char-
acters in the ship as a curiosity because he knows the usual trajectory
of the dropship: the commentary commences after a dash, the thresh-
old between the narration itself and the presentation of a general rule
based on the adverb invariably, the generic plural the archangel-class star-
ships and human beings. From this general rule, the narrator can draw
a conclusion by deduction at the beginning of the next paragraph: the
three characters are therefore not human. As Langlet (2006, 68) says,
“les romans en ‘je’ sont une reserve inépuisable de ce que Saint-Gelais
appelle le ‘didactisme honteux’, c’est-à-dire une articulation peu dis-
crète, ‘cousue de fil blanc’, du novum et de son explication” (the ‘I’
182
novels are an inexhaustible reserve of what Saint-Gelais calls ‘shame-

ful didacticism’, that is, an unobtrusive articulation, ‘sewn with white
thread’, of the novum and its explanation; our translation).
Fourthly, the explanatory segment may be assigned to a character
who instructs a less educated individual: “Plutôt que d’insérer dans le
récit des discours didactiques pour expliquer les données d’un problème
ou situer un lieu, les auteurs actuels préfèrent recourir aux dialogues”
(Bozetto 2007, 61) (Rather than inserting didactic discourses into the
narrative to explain the data of a problem or to locate a place, contem-
porary authors prefer making use of dialogues; our translation). It is
therefore no coincidence that of the 200 occurrences selected, 33.5% of
the prepositional phrases <PREP_definite article._ <FW>>appear in fic-
tional representations of speech (that is, direct speech: 21.5%, and indi-
rect or internal speech: 12%)—the rest being narrative. Fiction words
are therefore also found in fictional representations of speech, no matter
if it is their first appearance or not.
(31) Bon, soyons pratique: le danger, si danger il y a, réside dans la fusion

provoquée par la sexomorphose, parce qu’elle vous altère en profondeur
sans que vous puissiez percevoir les changements et que, à l’évidence, elle
vous uniformise. (Ayerdhal Sexomorphoses, 1994) (Well, let’s be practical:
the danger, if there is any danger, lies in the fusion caused by sexomorpho-
sis, because it alters you in depth without you being able to perceive the
changes and because, obviously, it makes you uniform.)
In this example, the speaker (a woman called Mesar) justifies her posi-
tion (where the danger comes from) in the causal subordinate clause.
The erudite character is either more experienced, as in this case, or has
expert knowledge, as in the following example:
(32) Prenez garde à vous: Point-Rouge est la plaque tournante des trafics
de l’index, en particulier du trafic du bétail humain, des esclaves. Ne
comptez pas sur les interliciers fédéraux pour vous aider en cas de pépin.
(Bordage Les Guerriers du silence, 1993) (Be careful: Point-Rouge is the
hub of the index traffic, especially the trafficking of human livestock,
slaves. Don’t rely on federal interliciers to help you in case of problems.)
183
The speaker is a travel agent who explains what he knows about Point-
Rouge to the woman intending to go there. The prepositional phrase sur
les interliciers fédéraux incorporates this didactic cautionary segment.
Finally, the prepositional phrase containing the fiction word may
appear in a narrative learning context. But, in the selected occurrences,
only the English corpus contains this type of example. In the following
excerpt, the speaker suggests an educational programme about the fic-
tion word:
(33) Each night, he began, I hope to continue our education on the goals
of this journey, to discuss the nature of the ecoi and their benefits and
potential dangers. (Bear Legacy, 1995)
The ship’s captain is giving his crew a little lesson. The speaker among
the characters who utters the explanatory speech is thus sometimes a
teacher. This is the case with the character named Hiroko, when he asks
his students a question:
(34) “All the names for Mars in the areophany are names given to it by
Terrans. About half of them mean fire star in the languages they come
from, but that is still a name from the outside. The question is, what is
Mars’s own name for itself?” (Robinson Green Mars, 1993)
And finally, the explanatory segment which includes the prepositional

phrases studied here may even appear outside the story:
(35) HOMERS. “Ceux qui rentrent à la maison.” Surnom donné à la frac-

tion de la population du Veld ayant choisi de se réinsérer au sein du Village.
Les concours de la fonction publique – bien que le nombre de postes à pour-
voir soit en chute libre depuis des décennies – constituent l’un des moyens les
plus couramment utilisés par les homers pour parvenir à leurs fins. (Lehman
F.A.U.S.T (1), 1996) (HOMERS. “Those who come home.” Nicknamed
after the fraction of the Veld population that has chosen to reintegrate
into the Village. Public service competitions – although the number of
positions to be filled has been falling dramatically for decades – are one of
the most common methods used by homers to achieve their goals.)
184
This definition can be found in the second appendix of the novel titled
“Lexique”. It is part of what Irène Langlet (2006, 105) calls “polytex-
tualité cloisonnée” (compartmentalized polytextuality; our translation),
that is, the non-narrative appendices of the novel, in this case the lex-
icon or glossary. In the example, the definition is provided as such, in
encyclopaedic form. The present continuous—soit, constituent—suggests
that the lexicon itself belongs to the fictional world, since the present
refers to the enunciation moment of the narrative.
4.3 Summary of the Uses of the Four Prepositional

Phrases <PREP_Definite Article_<FW>>
To begin with, it seems important to underline the lexical research

carried out by authors of science fiction (more so than in other sub-
genres most certainly). Indeed, <FW> pivots are not necessarily fiction
words: they can be uncommon or rare derivatives, borrowings from
foreign languages, scientific words belonging to different technolects
from diverse technology and natural science disciplines. The SF novel is
therefore based on an erudition that borders on creation (or intertwines
with it) to evoke the phenomenon of estrangement (Suvin 1979): the
reader is totally immersed in a different world, thanks to this “lexical
vertigo”. Science fiction is built on the technicality of its lexicon which
gives an impression of mastery (even if the plot happens to be confus-
ing), allowing the projection of an imaginary world.
To return to the four prepositional phrases chosen by us for their
specificity, it should be recalled that they are only samplings, 50 occur-
rences for each prepositional phrase <par le <FW>>, <sur le <FW>>, <of
the <FW>>, and <in the <FW>>. The present study suggests that
the works in the French corpus are based more on narrative than on
explanatory or descriptive sequences (that is, characters defined by their
actions expressed by agent phrases, <sur le <FW>> complementing verbs
and action nouns, and not, for example, adjuncts of place—whereas the
preposition sur can have a spatial meaning, like in ).
The four prepositional phrases all use fiction words representing
political systems, social roles occupied by characters, species (human or
analogous, animals, plants), and vessels carrying entire populations.
185
We might add that the prepositional phrases, essentially from the

English corpus, can show that the fiction words are integrated into cate-
gories (hypernymic relationships, aspectualization, singling out a group).
These categories make it possible to structure the fictional world in a
double movement of incorporation and singularization: to be distin-
guished from a whole, an individual (or an object, a machine, etc.) must
first be related to a category; one might even argue that the latter is cre-
ated by the author to highlight what the narrative sets apart from it.
The contexts in which the prepositional phrases are embedded allow
abductions, since the syntactic environment enables readers to formu-
late a hypothesis if the fiction word has not been defined yet (or has
been poorly defined), and then the syntactic context allows an abduc-
tion to be transformed into induction. Among the 200 occurrences
studied, we found many examples of didactic segments: in the form
of appositions (especially for French), explanatory contexts (expla-
nations provided by the heterodiegetic or homodiegetic narrator, or
by a particularly erudite character); the contexts seem more varied in
the English corpus. The samples may indicate that the works included
in the English corpus are more ambitious than those selected for the
French corpus (although this tentative conclusion may be biased by the
sampling). Indeed, history is written over generations (Simmons, Bear),
and the authors create worlds not only synchronously (in terms of their
spatial extension), but also diachronically (with respect to their tem-
poral dimension); moreover, it is especially in the English corpus that
metalinguistic reflections on the fiction words themselves are assigned to
the characters.
5 Conclusion
The present study, based on a fairly small corpus of 98 novels, shows
a certain convergence between American and French science fiction.
Although French SF is part of a specific French literary tradition,
described by Bréan (2012), which can be traced back to influential
authors like Jules Verne, the universe created by science fiction works in
American and French literary cultures has ended up merging in recent
186
decades. This fusion was perhaps inevitable given the strong influence
of American literature in the field and the huge number of transla-
tions from (American) English into French (which is not reciprocated).
Another factor of convergence may be found in what Damien Broderick
(1994, 57–60) calls the megatext, a concept that the Encyclopedia of
Science Fiction (SFE) defines as follows: “Science fiction is written in a
kind of code, a difficult vernacular learned through an apprenticeship.
Its decoding depends importantly on access to a megatext – the huge
body of established moves or reading protocols that the reader learns
through immersion in many hundreds of sf short stories and novels
(and, with significantly less sophistication, from movies, television epi-
sodes, and games). The sf megatext comprises a virtual encyclopaedia
and specialized dictionary”.10
The presence of “transfuges”, fiction words that cross the boundaries
of works by individual authors, but also the boundaries of languages
(such as holo, psionic, terraform, astrogation, conapt, cyborg, etc.), high-
lights the great unity in this genre-specific imaginary, despite the diver-
sity of inventions and findings.
To explain the behaviour of these fiction words from a phraseological
point of view, it should be noted that they are sometimes partially or
even totally transparent, which limits a priori the need for explanatory
intervention by the narrator. This (partial) transparency contributes to
placing the readers in a position of abduction, by forcing them to estab-
lish hypotheses based on the contextual or morphological clues given
to them. The marked tendency to insert fiction words in prepositional
phrases may play a similar role: giving some clues without breaking the
rhythm of the narration and relying on the reader to infer what is not
explicitly detailed. Indeed, in science fiction, the reader is supposed
to play an active part in the construction of the narration. Landragin
(2018, 144) calls this involvement the sense of reading, which completes
the more passive sense of wonder: “Le lecteur de SF doit ainsi passer
d’une illusion à une autre, d’une hypothèse à une autre, ce qui nécessite
un travail cognitif, c’est-à-dire une vigilance envers les indices qui sont
10http://www.sf-encyclopedia.com/entry/sf_megatext (consulted on January 2018).

187
autant d’éléments pouvant rejoindre sa xéno-encyclopédie. Ceci s’ajoute

sans se confondre au plaisir procuré par l’immersion dans un monde
imaginaire construit par l’auteur” (the SF reader must thus pass from
one illusion to another, from one hypothesis to another, which requires
cognitive work, that is, vigilance towards the clues, which are all ele-
ments that can join his xenoencyclopedia. This is added without com-
promising the pleasure provided by immersion in an imaginary world
built by the author; our translation).
Acknowledgements Many thanks to Judith Chambre, who helped in the

selection of the English and French novels of the corpus.
References
Adam, Jean-Michel. 2017 (1992). Les Textes: types et prototypes. Paris: Armand
Colin.
Angenot, Marc. 1978. “Le paradigme absent. Éléments pour une sémiotique
de la science-fiction.” Poétique 33 (2): 74–89.
Bordas, Éric. 2001. “Un style dix-neuviémiste: le determinant discontinu un
de ces… qui.” L’Information grammaticale 90: 32–43.
Bozetto, Roger. 2007. La Science-fiction. Paris: Armand Colin.
Bréan, Simon. 2012. La Science-fiction en France: théorie et histoire d’une littéra-
ture. Paris: PUPS.
Broderick, Damien. 1994. Reading by Starlight: Postmodern Science Fiction.
London: Routledge.
Chambre, Judith, and Olivier Kraif. 2017. “Identification de traits spéci-
fiques du roman policier et de science-fiction.” In JLC 2017, 9èmes
Journées Internationales de la Linguistique de corpus, 3–6 July 2017,
Grenoble. https://hal.archives-ouvertes.fr/hal-01884804.
Csicsery-Ronay, Jr., Istvan. 2013. “Que voulons-nous dire quand nous parlons
de ‘science-fiction mondiale’.” Translated by Samuel Minne. ReS Futurae 3.
http://journals.openedition.org/resf/411.
Kraif, Olivier, and Julie Sorba. 2018. “Spécificités des expressions spatiales
et temporelles dans quatre sous-genres romanesques (policier, science-
fiction, historique et littérature générale).” In JATD’18 Proceedings of the 14th
International Conference on Statistical Analysis of Textual Data, vol. 1, 392–
399, 12–15 June, Rome. https://hal.archives-ouvertes.fr/hal-01844460.
188
Kraif, Olivier, and Sascha Diwersy. 2012. “Le Lexicoscope: un outil pour l’étude
de profils combinatoires et l’extraction de constructions lexico-syntaxiques.”
In Actes de la conférence TALN 2012, 399–406, 4–8 June, Grenoble. http://
www.aclweb.org/anthology/F12-2033.
Landragin, Frédéric. 2018. Comment parler à un Alien? Langage et linguistique
dans la science-fiction. Saint-Mammès: Éditions du Bélial.
Langlet, Irène. 2006. La Science-fiction. Lecture et poétique d’un genre littéraire.
Paris: Armand Colin.
Poix, Cécile. 2018. “Neology in Children’s Literature: A Typology of
Occasionalisms.” Lexis 12. http://journals.openedition.org/lexis/2111.
Prucher, Jeff (ed.). 2017. Brave New Words: The Oxford Dictionary of Science
Fiction. Oxford: Oxford University Press.
Rolls, Heather, and Michael P.H. Rodgers. 2017. “Science-Specific Technical
Vocabulary in Science Fiction-Fantasy Texts: A Case for ‘Language Through
Literature’.” English for Specific Purposes 48 (October): 44–56.
Saint-Gelais, Richard. 1999. L’Empire du pseudo: Modernités de la Science-
fiction. Montreal, QC: Nota Bene.
Suvin, Darko. 1979. Metamorphoses of Science Fiction: On the Poetics and
History of a Literary Genre. New Haven and London: Yale University Press.
Annex
Authors in the French corpus: Ayerdhal, Barberi, Berthelot, Bordage, Brussolo,
Colin and Gaborit, Curval, Dantec, Deff, Di Rollo, Dunyach, Fontana,
Genefort, Lehman, Léourier, Ligny, Pagel, Pelot, Wagner, Walther, Werber.
Authors in the English corpus: Asaro, Asimov, Barnes, Bear, Brin, Bujold,
Butler, Cherryh, Crichton, Goonan, Haldeman, Kress, McCaffrey, Moon,
Pohl, Robinson, Scott, Silverberg, Simmons, Stephenson, Gibson and
Sterling, Sterling, Swanwick, Vinge, Williams, Willis.
7
Science Fiction versus Fantasy: A Semantic
Categorization and its Contribution
to Distinguishing Two Literary Genres
Vannina Goossens, Clémence Jacquot
and Susanne Dyka
1 Introduction
This chapter proposes a methodological reflection on the analysis and
classification of recurrent lexico-syntactic trees (RLTs, see Chapter 2
by Legallois and Koch in this volume) extracted from modern narrative
corpora. We will show how those patterns can be used for defining and
delimiting particular fictional subgenres in a contrastive approach. We
will focus on the contribution of a semantic classification of the pat-
terns that were extracted from the corpora in the PhraseoRom project.
V. Goossens (*)
C. Jacquot
University Paul-Valéry Montpellier 3, Montpellier, France
e-mail: clemence.jacquot@univ-montp3.fr
S. Dyka
e-mail: susanne.dyka@fau.de
https://doi.org/10.1007/978-3-030-23744-8_7
190
V. Goossens et al.
This classification allows us to compare two closely related narrative

subgenres: science fiction and fantasy.
Both fantasy and science fiction belong to the same editorial cate-
gory the French call “littérature de l’imaginaire” (speculative fiction).1
This mixed-genre category shows us that the two are sometimes dif-
ficult to tell apart, causing some specialists to argue about their liter-
ary, thematic and structural differences. Besson (2007, 39) or Baudou
(2005, 58), for example, discuss what sets these two genres apart and
the porosity of the boundaries between them. Many literary experts
also underline the increasing hybridization of the two genres (see
Bazin 2015, 111) especially during recent decades. It affects, for exam-
ple, subcategories of science fiction and fantasy like uchrony novels
(alternative history novels in which a variation of an actual historical
event is imagined), steampunk novels (whose plot takes place in the
Victorian age of steam) or science fantasy novels (which mix science
fiction and heroic fantasy by using modern technologies in an ancient
or medieval world).
This hybridization process is also intensified by attaching cer-
tain editorial labels which are sometimes just marketing ploys. This
is exemplified by some novels when they are published several times
and are relabeled (as science fiction or fantasy) between printings.
For example, Francis Berthelot’s Khanaor, a two-volume cycle of
heroic fantasy, was published in 1983 as a fantasy novel—only to be
republished in 2010 under the label science fiction in the Folio-SF
collection.2
1This label is, as we said, a French editorial category which includes science fiction, fantasy and
a part of the French category called “fantastique” (see Hommel 2017, 9; Torres 1997). In the
English editorial field, we also found the label “speculative fiction” including for example science
fiction, fantasy and superhero fictions. These concepts and vocabulary are indeed used a bit dif-
ferently by French and English critics. In any event, the basis of our theoretical framework is
French. See Cornillon (2012, 15 ff).
2The Folio-SF collection presents stories and novels which take place in many different possible
imaginary worlds. It includes several French literary genres such as “fantastique”, science fiction
and fantasy, but also unclassifiable imaginary subgenres which mix these categories. See Folio-SF
collection’s presentation here: http://www.folio-lesite.fr/SF-Fantasy.
7 Science Fiction versus Fantasy: A Semantic Categorization …
191
Our approach is based on a semantic categorization and addresses

these literary questions by providing corpus-based data that are rele-
vant in stylistics. We will first outline the methodology and describe the
semantic categorization of the specific RLTs extracted from our corpora.
Then we proceed to discussing some stylistic distinctions, which allows
us to show linguistic similarities and differences between the two genres
of science fiction and fantasy.
Comparing these two genres here will also let us propose a broader reflec-
tion on the interpretative stance underlying the semantic classification of
our corpus and its methodological implications. It raises these questions:
What kind of results can the classification of RLTs with a grid yield? How
identical are the profiles identified in English and French? But also: Do
these categories provide food for thought for interpreting texts? To what
extent can this annotation be designed to describe and highlight convincing
results from the different narrative subgenres of the PhraseoRom corpus?
2 Corpora and Methodology

Our analysis is based on the corpora created in the PhraseoRom project.
The corpora of fantasy and science fiction are comparable, in French
and in English as shown in Table 1 (see also Appendix A).
2.1 The French Corpora
Our French corpus of science fiction novels includes among others

authors like Y. Ayerdhal, R. Barjavel, P. Bordage, M. Dantec, and R.
Merle. This corpus covers novels from 1952 to 2014 and thus integrates
Table 1 French and English science fiction and fantasy corpora

FR EN
SF FY SF FY
Tokens 13,173,618 13,323,976 12,962,491 17,814,599
Authors 39 43 39 41
Books 147 104 116 142
192
V. Goossens et al.
several subgenres of science fiction novels: utopian novels like G. Klein’s

Les Seigneurs de la guerre, anticipation novels like Y. Ayerdhal’s Cybione
cycle, planet opera like L. Genefort’s Omale cycle and dystopia like P.
Boulle’s La Planète des singes.
The French fantasy corpus contains authors like Ange, F. Berthelot,
Ch. Bousquet, J.-L. Fetjaine, M. Gaborit, J.-Ph. Jaworski, A. Robillard
and P. Pevel. This corpus includes an important range of novels in con-
temporary fantasy subcategories: the steampunk universe (for Albert’s
Avant le Déluge ), heroic fantasy (such as Berthelot’s Khanaor ), dark
fantasy (such as Bousquet’s Arachnae ) and Arthurian fantasy (such as
L’Apprenti de Merlin by Clavel).
2.2 The English Corpora
The English corpus of science fiction novels is reasonably comparable

to the French science fiction corpus. It includes one of the most impor-
tant references of post-Second World War British science fiction nov-
els: E.C. Tubb and his epic SF cycle The Dumarest Saga. It also features
several well-known prolific authors such as A. C. Clarke and S. Baxter,
both representing the hard SF subgenre. Our corpus illustrates several
trends in science fiction, for example, dystopian novels like Burgess’s
The Clockwork Orange or the cycles and novels of J. G. Ballard, K. Amis,
and B. Aldiss. Also found in it are some examples of space opera (I. M.
Banks or P. F. Hamilton’s novels) and of “comedy” science fiction novels
(that is, writings which play with traditional science fiction conventions
by using satire and humor about the genre itself or about situations of
the present society), such as the H2G2 cycle by D. Adams.
The English corpus of fantasy novels integrates some of the greatest,
most internationally representative authors in this genre. Their success
is both economic and symbolic, since some of them such as J. R. R.
Tolkien and his The Lord of the Rings cycle—which represents a model
of high fantasy novels—and C. S. Lewis (The Chronicles of Narnia )
are mentioned as references (especially by French writers). The corpus
also includes J. K. Rowling’s Harry Potter cycle (low fantasy novels)
193
and T. Pratchett’s mixed heroic fantasy-cum-humor novels. Some other

famous names and their iconic works are Ph. Pullman and His Dark
Materials cycle (high fantasy also cited as science fantasy novels), E.
Colfer’s Artemis Fowl cycle, R. Dahl’s Charlie and the Chocolate Factory,
N. Gaiman’s Coraline (a dark fantasy reference) or C. Barker’s dystopian
novels.
2.3 Extraction of Recurrent Lexico-Syntactic

Trees (RLTs)
The RLTs were extracted from the science fiction and fantasy corpora
by the method used in the PhraseoRom project outlined in Chapter 2
by Legallois and Koch in this volume. We used a dispersion above or
equal to 10 authors for selecting patterns specific to a subgenre and not
just to a single author (a high level of dispersion provides more generic
patterns). We added a morphological criterion by limiting ourselves to
patterns implying a verb. Table 2 shows the specific pattern totals for
each genre and the number of patterns we generated after applying our
thresholds.
The table shows that there are fewer RLTs in English than in French.
We arrived at similar results for the other subgenres studied in the
PhraseoRom project, indicating that this phenomenon is not just spe-
cific to the two genres studied here. Nor is it related to a specificity of
the English language; more likely it stems from the differences in syn-
tactic annotation (which is the basis of the lexico-syntactic patterns)
Table 2 Cumulative thresholds and number of RLTs for each language and
genre
English French
SF FY SF FY
LLR ≥ 10,83 3731 5070 6274 7097
Verb 920 2006 2474 4029
Dispersion ≥ 10 154 174 405 864
194
V. Goossens et al.
between the two languages. We chose not to raise the number of RLTs
by lowering the dispersion for the English corpus because this criterion
significantly impacts the structures identified: the more they are pres-
ent in a large number of authors, the more generic they are, and vice
versa. We realize that such a difference in the number of patterns ana-
lyzed between French and English can lead to non-significant quantita-
tive differences but accept it because we do not propose a quantitative
analysis here. It also bears stressing that the total number of RLTs does
not matter because it is a highly redundant mass of data; moreover, all
RLTs that revolve around a certain collocation are only a guide for iden-
tifying patterns.
We also extracted fewer RLTs in science fiction than in fantasy
in both languages even though the gap is more striking for French.
The difference may be attributable to the lexical characteristics of
science fiction, which employs many neologisms that in the ensem-
ble constitute a xenoencyclopedia (see Chapter 6 by Gonon and
Kraif in this volume). These fiction words occur too infrequently
and, above all, are specific to authors and not to genres: they did
not pass our dispersion thresholds and therefore did not enter the
specific RLTs.
For categorizing the large number of patterns extracted for each
genre in the project, we settled on applying a semantic grid to the list of
extracted RLTs as our preferred methodological approach.
3 Presentation of the Semantic Grid

As mentioned in the previous section, the prerequisite for develop-
ing the grid was that the RLTs we would work with had to contain a
verb. This lets us reduce the amount of data to manageable levels while
excluding RLTs corresponding to strictly referential entities that were
not really analyzable from a semantic point of view. But, it also made
for better coherence with respect to the complexity of the patterns and
guided our choice of semantic dimensions.
195
Table 3 Semantic dimensions Dimensions Values

and values used in the
Action Undefined
PhraseoRom project Movement
Travel
Other
Communication Undefined
Verbal
Physical
State Undefined
Animate
Inanimate
Abstract
Cognition
Qualia Undefined
Sensation
Perception
Affect
Time
Place
Event
Other
–
As shown in Table 3, we developed a semantic classification com-

prising eight dimensions3 and up to four values for each dimension.
Additionally, we set up the category “other” for unclassifiable cases and
a category (“-”) for filtering out the RLTs that did not fit the criteria
and thus constituted noise, for example, in cases of bad syntactic anno-
tation. All our multi-value dimensions have one value (“undefined”) in
common out of necessity because in some cases we were unsure which
of the other values in a dimension applied: the “undefined” value thus
became the default option for these ambiguous situations. The pattern
<was in the air>, for example, could not be analyzed as either “state:an-
imate” or “state:inanimate”—both in fact are possible (and do occur in
our corpus), hence it was analyzed as “state:undefined”.
3See Mahlberg (2007, 228), who uses functional groups to classify her findings.
196
V. Goossens et al.
If we encountered a pattern with a complex meaning, we permitted

a two-fold classification: for example <lever les yeux au ciel>4 (look up to
the sky) is categorized as “action:other” and “qualia:affect”. A dual cate-
gorization was also acceptable for verbs that belonged into one meaning
dimension, like follow in <follow into the kitchen> where its complement
into the kitchen fit into a different category. This particular pattern we
analyzed as “action:travel” and “place”.
In cases of polysemy, a pattern’s possible meanings are also assigned
to two categories. For instance, <have a look> was analyzed as
“action:other” or “state:undefined” as it could either be someone having a
look at something or something having a look (for example a face). <Go the
wrong way> illustrates another potential two-fold categorization. Going
the wrong way as in choosing the wrong path is analyzed as “action:travel”,
whereas if something goes the wrong way, as when something gets stuck
in the wind pipe, it is tagged as “event”. A similar case in French is
<faire le tour de> as “action:travel” or as “cognition”.
3.1 “Action” and “Communication”
The “action” dimension comprises all actions performed by an agent in

a dynamic situation. The category contains the four values of “unde-
fined”, “movement”, “travel” and “other”.
– “Action:movement” describes actions that imply some sort of move-

ment without changing place horizontally (vertical movement). This
includes movements with the whole body (<rise to his feet>), move-
ments toward another person (<give a hug>), movements of or with
body parts (<hold his hands up>) and movements of the face or part
of the face (<open his eyes>, <wiped his brow>, <smile at the thought>).
Often these cases have a communicative aspect, which is considered
in our analysis by adding a second category “communication”. Cases
such as <poing s’abattre> (“sur la table” for example) (the fist hits the
197
table) or <sa tête s’incline> (his head bows), where the body part is
the subject, were categorized as “event” (see Sect. 3.2). With verbs
denoting an action as well as a state, we focused on the semantics of
the verb and classified it as both: for example <sit down in armchair>
was classified as “action:movement” and “state” but <was sitting in an
armchair> as “state”.
– “Action:travel” implies movement from a point A to a point B (on
a horizontal axis), even if the two points are not made explicit.
These include patterns like <walk down the aisle>, <burst into the
room>, <leave the room>. “Place” very often is indicated as a second
category.
– “Action:other” captures all actions not indicating (body) move-
ment but also a movement along the horizontal axis. The category
includes a vast variety of actions: <accept the offer>, <blow out the
candle>, <broke the connection> (and “communication”), <close the
curtains>, <make a cup of coffee>, <have a drink>, <avoid like the
plague>, <try to avoid>, <look at the camera>, <follow his gaze>. This
huge category calls for more detailed analysis to identify subcatego-
ries that could be important for one genre without being relevant
for all the subgenres (see Sect. 4.1 on the warlike actions specific to
French fantasy).
Into the communication category with its three values of “undefined”,

“verbal” and “physical” fall patterns that express either the intention
to, or the act of, sharing information, feelings, thoughts, et cetera, with
another person:
– “Communication:verbal” receives patterns that imply using the artic-

ulatory organs for communicating regardless of whether intentional
or not. Possible entries under this category include <have a chat>,
<give out a harsh shout>, <clear his throat>. Sounds made by animals
are also categorized as verbal communication.
– “Communication:physical” is reserved for patterns that express
non-verbal contacts initiated for sharing information, feelings,
thoughts, et cetera, with another person (physically), for example
<acknowledge with a nod > or <manage a weak smile>.
198
V. Goossens et al.
3.2 “State” and “Event”
The “state” dimension has four values (“undefined”, “animate”,

“inanimate”, “abstract”) and mainly describes states of animate, inan-
imate and abstract agents. The state might be permanent, for exam-
ple <have blond hair>, or it might be temporary, for example <was in
danger>.
– “State:inanimate” applies to states of an inanimate object, for exam-

ple <made of silk>.
– “State:animate” refers to animate subjects, like <she was wearing a
dress> or <he has a moustache>, or to the description of personality
traits like <he was charming>.
– “State:abstract” classifies patterns that include an abstract subject like
<la situation est grave> (the situation is serious).
Weather phenomena which are static—<the sun is in the sky> for

example—are classified as “state:inanimate”. Changes of states like cer-
tain other weather phenomena (<the rain was falling>, <the sun was ris-
ing>, <the sun was setting>) on the other hand would be categorized as
“event”. “Event” as a dimension describes dynamic processes or actions
without an agent, such as <a cheer went up>, <the baby was born>,
<appear on the screen>.
3.3 “Cognition” and “Qualia”
“Cognition”, a category devoid of values, is about gaining or los-

ing knowledge.5 Belonging in this category are RLTs like <think for a
moment> (and “time”) and <have no doubt>.
The “qualia” category (with the values of “undefined”, “sensation”,
“perception” and “affect”) is reserved for feelings, emotions and their
perception.
5The Oxford Dictionary Online defines it as: “The mental action or process of acquiring knowl-
edge and understanding through thought, experience, and the senses.”

199
– “Qualia:sensation” accommodates feelings that are experienced phys-

ically or perceived with the body, as for example <feel the pain>, <feel
the warmth> or <have a headache>.
– “Qualia:perception” is designed for mental representations of a sensation.
It includes activities by which a subject or being experiences objects and
activities based on inputs from the sense organs: seeing, hearing, touching,
tasting, smelling, for example <hear on the phone> or <have a taste>.
– “Qualia:affect” categorizes patterns involving feelings and emotions as
well as psychological states which vary in their intensity and are gener-
ally experienced as positive or negative by an animate subject (being).
They may be connected to a source (a cause or an object), inserted in
a temporal frame and incapable of being observed directly: <be in love
with someone>, <she was delighted>, <they are scared>.
3.4 “Place” and “Time”
The category “place” classifies a place or location. Objects (for exam-

ple pieces of furniture: <was in armchair>) are mostly excluded, except
when they imply a place, for example, doors and windows when they
are used with a verb and a preposition of location to imply a place
(<move from the window> but not <open the window>). “Place” fre-
quently is a second category, as for example in <follow into the kitchen>,
<he appeared in the doorway> or <was on the beach>.
“Time” categorizes patterns which are about time or have at least a
temporal component: <it was in the afternoon>, <be in the eighties>. In
the latter case they are classified as “time” as a second category: <think
for a moment>, <the sun was sinking>.
3.5 “Other”
The category “other” is a mixed bag category of patterns without an

identified suitable category, even though they may be of interest other-
wise. This often includes highly polysemous patterns, for example, NP +
is which could be classified as a state (<is nice>), an action (<is running>)
or a qualia (<is feeling sick>). Idiomatic and other fixed expressions like
200
V. Goossens et al.
<what the hell are you doing> find a home in this category. We will dis-
cuss the content of the category “other” for English science fiction in
the next section.
It bears repeating here that we developed this semantic classification
on the basis of our data not as a theoretical exercise but as a practical tool
for an initial categorization of the large number of patterns we extracted
during the project. Like any annotation, this categorization necessarily
emphasizes certain semantic characteristics of our patterns (actions for
example, important in analyzing fiction) while it deemphasizes others.
Simply put, the grid had to remain manageable, given that the RLTs
had to be annotated manually.6 The RLTs we extracted from our corpus
are first in the queue as the project team begins the semantic analysis.
Categorizing them necessarily precedes the identification and analysis of
motifs (see Chapter 2 by Legallois and Koch in this volume) with discur-
sive functions (see Appendix B for definitions and Chapter 8 by Sorba
et al. in this volume for an exemplary analysis) and a second phase of
semantic description. In the following section, we will show how this
first rough semantic categorization can serve as a relevant space for nego-
tiation between disciplines in a project like ours and thus provide a basis
for further analysis of specific patterns in contemporary literature.
3.6 Application of the Semantic Grid to Science Fiction

and Fantasy in French and in English
We had the semantic grid manually applied to the French and English
extractions (described in Table 2) by three annotators and then dou-
ble-checked by one of them. As Table 4 shows, for the semantic clas-
sification of French patterns7 some categories appear more frequently
6We are currently working on the development of a script which will allow semi-automatic annota-
tion of RLTs based on the first annotations performed on genre pairs and using a vector analysis of
the similarity of our RLTs.
7We note that the total number of semantic dimensions is higher than the number of extracted
RLTs presented in Table 2: this is due to the double categorization of some RLTs presented in
Sect. 3.
201
Table 4 Classification of patterns by semantic dimension for French science fic-

tion and fantasy
'LPHQVLRQV 9DOXHV 6FLHQFHILFWLRQ )DQWDV\
7RWDO 7RWDO
1% 1%
$FWLRQ 8QGHILQHG
0RYHPHQW
7UDYHO
2WKHU
(YHQW
6WDWH 8QGHILQHG
$QLPDWH
,QDQLPDWH
$EVWUDFW
4XDOLD 8QGHILQHG
$IIHFW
6HQVDWLRQ
3HUFHSWLRQ
&RJQLWLRQ
7LPH
3ODFH
&RPPXQLFDWLRQ 8QGHILQHG
3K\VLFDO
9HUEDO
2WKHU

7RWDO
relative to some genres than others. While some dimensions—

“action:other” (although in greater numbers in fantasy than in science
fiction), “event” and “communication:verbal”—occur quite frequently
for both genres (highlighted in green in the table), other categories
are over-represented in one genre as opposed to the other, namely:
“action:movement” and “action:travel” for fantasy (highlighted in yel-
low) and “cognition” for science fiction (in pink). Finally, between
202
V. Goossens et al.
10–15% of patterns wound up in “other” or “-”. We found similar dis-

tributions of those two last categories for other pairs of genres (see, for
example, Goossens et al. 2018; Gonon et al. 2018 for application of a
previous version of the semantic grid).
The results for the English corpus shown in Table 5 turned out sim-
ilar to those for the French corpus albeit with slight differences: the
Table 5 Classification of RLTs by semantic dimension for English science fiction

and fantasy
'LPHQVLRQV 9DOXHV 6FLHQFHILFWLRQ )DQWDV\
7RWDO 7RWDO
1% 1%
$FWLRQ 8QGHILQHG
0RYHPHQW
7UDYHO
2WKHU
(YHQW
6WDWH 8QGHILQHG
$QLPDWH
,QDQLPDWH
$EVWUDFW
4XDOLD 8QGHILQHG
$IIHFW
6HQVDWLRQ
3HUFHSWLRQ
&RJQLWLRQ
3ODFH
7LPH
&RPPXQLFDWLRQ 8QGHILQHG
3K\VLFDO
9HUEDO
2WKHU

7RWDO
203
“action:other” category, while frequently occurring for both genres, as

in French, did so more often for fantasy. The “event” category is rarer
(but, as in French, nearly identical for fantasy and science fiction); how-
ever, the incidence of “state” is quite high for both science fiction and
fantasy (almost always with “place” as a second category). The “place”
category, almost non-existent in French, is prevalent in English. As in
French, the “action:movement” category is highly specific to fantasy in
English but “action:travel” is not important. “Communication:verbal”,
a common dimension for French, in English is specific to fantasy. As
in French, “cognition” is the only over-represented dimension for sci-
ence fiction. The most voluminous category for English science fiction
is “other”, capturing more than 38% of RLTs. This is a much higher
proportion than was found for both French science fiction and other
genres in previous studies. We will have more to say about RLTs in this
category in Sect. 4.1.3.
We follow this overview of semantic dimension classifications with a
contrastive approach to demonstrate the grid’s potential contribution to
distinguishing science fiction from fantasy. The two tables above seem
to show that the major differences exist between science fiction and fan-
tasy rather than between French and English. In the next sections, we
will propose both a stylistic analysis of these genre differences and dis-
cuss what the classification of RLTs with our semantic grid revealed.
4 Contribution to Differentiating
Science Fiction and Fantasy
in French and in English
In our introduction, we labeled the science fiction and fantasy genres
as belonging to the “littérature de l’imaginaire”, an editorial category
that emphasizes the similarities of novels whose plot and characters help
construct an imaginary world. But there are also important theoretical
distinctions between these two genres, for example, in the way each of
them represents the world.
204
V. Goossens et al.
The traditional elements invoked to establish distinctions within

the massive and disparate corpus are often simplistic. The main differ-
ence claimed is a conflict between the science in science fiction and the
magic in fantasy. Also implied is the perhaps more convenient opposi-
tion between rational sense on the one hand and imagination on the
other that Besson calls “régimes de sens” (that is, opposing schemes for
representing reality and its comprehension, Besson 2007). But, more
recent literary analysis points out that the difference may instead lie in
the way we read the genres (Saint-Gelais 1999). This difference would
seem to let us examine the specificities of each of these genres most
effectively. Saint-Gelais explained that science fiction and fantasy offer
readers opposite experiences: where science fiction initiates awareness
or attentiveness in the reader’s mind, fantasy books play with his con-
fidence and his euphoria. For instance, while a science fiction reader
is often on the look-out for clues to understand and discuss the scien-
tific explanation of a story based on the strange and unusual, a reader
of fantasy is supposed to accept magic and supernatural events as
factual. The reader must therefore trust the writer’s fiction as a valid
experience.
For this type of analysis, linguistic and stylistic criteria seem to
be somewhat secondary despite the fact that lexical creativity is really
important in both of these genres. Obviously, science fiction novels are
expected to resort much more to scientific expressions, while fantasy
novels are more likely to draw from mythical and historical references
(medieval ones, for example, in heroic fantasy). Hence, thematic ele-
ments could really be discriminative here and some lexical studies (Eco
1979; Saint-Gelais 1999, see also Chapter 6 by Gonon and Kraif in this
volume) show more precisely which xenoencyclopedia is used in each
case.
We assume here that our phraseological approach will allow us to go
beyond a lexical study by highlighting recurrent RLTs specific to several
different novels of the same genre.
In the course of our semantic annotation, several important catego-
ries emerged as specific to fantasy or science fiction. Therefore, we can
draw a distinction between results that we called generic markers and
other results that were less expected than those.
205
4.1 Generic Markers
Because they belong to popular fiction, science fiction and especially

fantasy are often considered as more stereotypical than what is normally
understood by literary fiction. Stereotyped structure and form are fre-
quently associated with popular novels. They are considered to be part
of “successful recipes”, and it would probably be interesting to discuss
these stereotypical ideas and expressions given the importance ascribed
to building a reader’s encyclopedia (see, for example, Baroni on crime
novels, 2003, 145).
But, our aim here is not to draw out clichés about these two gen-
res, rather it is to analyze how the phraseological approach based on the
semantic grid allows us to compare science fiction and fantasy by using
linguistic criteria.
We therefore define generic markers as RLTs that are specific to sci-
ence fiction or fantasy, that are expected to be specific, and furthermore
are included in certain topical actions or referential description that
immediately allow us to distinguish between the two genres.
4.1.1 Traveling in Science Fiction and Fantasy
The fantasy corpus in French especially is characterized by a greater

number of RLTs than the science fiction corpus. These RLTs cover a
wide variety of actions, both in French and English, and particularly in
the French “action:travel” category, where they account for 13% of the
RLTs whereas with 2% this category is almost absent in English. These
types of actions heavily involve prototypical situations and places: many
relevant RLTs, at times suitable for grouping as motifs, signal travel
through or within a specific landscape like forests, woods, occasional
castles or battlefields, et cetera. For instance, as in this motif:
– s’enfoncer | sortir | pénétrer dans la forêt | les bois (to disappear into | to
leave | to enter the forest | the woods)
The forest appears here as a particularly wild, mysterious or frightening

place; the functions performed by these motifs bear comparison with
206
V. Goossens et al.
certain contrasting actions that characterize leaving a city (RLTs like

<quitter la cité>, <sortir de la ville> leave the city). The French corpus
also points to the importance of the different stages of the travel and
includes mentions of trips to an inn (l’auberge the hostel) or to a room
(la chambre the room). However, this contrasting travel through the
woods and through cities is not evident in the English corpus. Here we
find <go to the kitchen> as the only specific RLT showing a character’s
movement. Just like the inn and the room, it serves as a quiet interlude
and, in this latter instance, as a break when people prepare something to
eat.
Several RLTs point to another specific locational aspect: the French
corpus often shows specific movements on stairs (both up and down). It
serves especially to underline the characters’ haste and excitement:
– grimper | descendre | dévaler | les marches | escaliers (climbing | going

down | running down steps | stairs)
Similarly, from the English corpus we extracted the motif walk up |

climb the steps | stairs (with three RLTs out of four in this category).
The “action:travel” category is much less present in the French sci-
ence fiction corpus (5%) and, once again, nearly absent from the
English one (5 RLTs representing less than 3%). In the French corpus,
the RLTs point in a very different direction: people are crossing the sky,
moving into space, or traveling in time (traverser | flotter | voyager dans
l’espace | le temps, to cross | to float | to travel in space | time). Science
fiction characters also set a course for new destinations: <mettre le cap>
(set a course), <mettre en orbite> (put into orbit).
The English corpus includes one thematic generic marker (<leave the
ship>) but, once again, the RLTs here are not as varied as in the French
corpus and therefore play less of a role. Various “action:travel” items
focus especially on speed: <move at speed>, <run the length > or <slow to
a halt>.
However, understanding these less diversified results in the English
corpus requires analyzing another important category: “place”. A con-
siderable number of the English corpus RLTs code as “place” both for
fantasy (7%) and science fiction (8%). In French, these RLTs only
207
represent 3% of science fiction and 1% of fantasy RLTs. This label

almost always appears in tandem with a second category, mainly “state”
(<be on the ship > for example). Although it is not the main category
(the place not being conveyed by the verb), it plays an important role in
our two genres as a generic marker. It serves to emphasize in particular
the landscapes we previously described (for example woods, mountains
and roads for fantasy), which is characteristic of science fiction. The
xenoencyclopedia described by Saint-Gelais plays a major role here as
the following motif shows: be in | on orbit | cabin | system | galaxy | uni-
verse | board | the surface | planet | the ship.
4.1.2 “Action:movement” as a Category for Distinguishing

Between Genres and Between Languages
At first glance, this category seems to be distributed in a similar way

in the two genres in both French and English: it constituted 12% of
the RLTs in fantasy (in French and in English), 3% (English) and 4%
(French) for science fiction. But, on closer examination, we realize that
once again the French corpus presents a large number of very specific
RLTs; especially striking is the over-representation of aggressive and
warlike movements, from the hostile RLT <serrer le poing> (to clench
your fist), to <poser sur le pommeau> putting [the hand] on the sword
hilt), <il se met en garde>, <baisser pour éviter> and <esquiver l’attaque>
(he takes up a defensive stance, ducking to avoid [something] and dodg-
ing the attack).
These kinds of movements are quantitatively as well as qualitatively
relevant in this context, and we can consider them as generic markers.
However, we exclude the English fantasy corpus from this because there
they did not emerge as RLTs.
Similarly, many movements in the French fantasy corpus express def-
erence or submission (whether granted or imposed). Their expression
involves several actions of the gaze direction, such as baisser les yeux | le
regard (look down) and movements implying more or less ritualized ges-
tures of prostration: <faire une révérence> (curtsy), <s’incliner devant lui>
208
V. Goossens et al.
(to bow down before him), <il s’agenouille> (to kneel [before]), baisser |
incliner la tête (to bow the head).
In the English corpus, such acts of submission are not specific, as we
found earlier in the case of warlike actions. The only exceptions may be
the two RLTs <fell to his knees> and <go to his feet>. Most of the specific
RLTs for English fantasy are parts of interactions and are highly expres-
sive, especially since they imply movements of the face (eyes, nose, lips,
teeth, etc.). They express contentment (to clap | rub his hands (together) ),
enthusiasm or empathy (<slap on the back> often followed by cheerfully ),
amazement (<roll his eyes>, <raise an eyebrow>), disappointment and
pain (<shake his head sadly>) or bitterness (<grit his teeth>).
These kinds of expressive movements are also reflected in the French
RLTs but our results indicate that the French corpus offers more intense
and varied examples than the English corpus. Therefore, this category
appears as more generically marked in French than in English.
To conclude, the results in English fantasy are quite comparable to
the science fiction corpora, both in English and in French. They mostly
share features, such as movements to reinforce interactions between two
characters, for instance, <purse his lips> or <tilt his head> in the English
science fiction corpus and, in the French science fiction corpus, <hausser
les épaules> (to shrug one’s shoulders), <je secoue la tête> (I shake my
head), <il se tourne vers moi> (he turns to me) and <me tourne le dos>
(turns his back on me) and actions to sit down somewhere.
4.1.3 “Action:other” and “State”: Two Other Categories

Supporting the Theory of Generic Markers
These two categories can also be analyzed as a pair because they contain
a large number of RLTs that converge in building a typical universe.
In effect, they combine various subcategories of RLTs, several of which
qualify as generic markers in both French and English.
In the fantasy corpus we observed numerous other RLTs related to
war activity. The “action:other” category contains a lot of fight actions
that form several motifs like dégainer | sortir | tirer | lever | lâcher son épée
(to draw a sword) and trancher | briser la gorge | la nuque | le cou | les os
209
(slit | break the throat | neck | bones). We also find some RLTs evoking
the effects of the fights in “state”: être couvert | maculé | éclaboussé de
sang (be covered | stained | splashed with blood), <cribler de flèches> (to
riddle with arrows). These war activities are also very well represented in
the “event” category (over-represented for fantasy and also for science
fiction, but only in French): la lame | flèche s’enfonce | s’abat | tranche |
se fiche (the blade | arrow goes in | slices | crashes | embeds itself in), <le
sang coule> (blood is flowing), <se vider de son sang> (to bleed out), et
cetera.
A second set of interesting RLTs, emerging from the latter two cate-
gories in the French corpus, designate a set of characters related to royal
or religious institutions and functions: être le roi | la reine | le fils | l’Élu
| le Seigneur | l’héritière and the like (to be the king | the queen | the son
| the chosen one | the Lord | the heir). This set of RLTs, which relate to
characters typical of fantasy, could be expanded if we look at the cate-
gory “other” which includes a large number of semantically unclassifi-
able RLTs of the type la sorcière | le mage | la princesse | le chevalier | la
souveraine est | a (the witch | the magician | the princess | the knight |
the sovereign is | has). In this category, we also find RLTs containing
topical clues such as dragons or monsters. Finally, the “action:other”
category offers us descriptive elements related to clothing (<porter à la
ceinture>, to wear on the belt, porter une armure | tunique | robe, to wear
armor | a tunic | a dress) that also tend to be quite typical.
In the fantasy English corpus, some RLTs in “action:other” offer up a
medieval vocabulary: <draw his sword>, <cut his throat>, <light the can-
dles>, <open the gate> or <stare into fire>. The “state” category does not
feature social or religious positions; all the states in the English corpus,
both in FY and SF, indicate geographical positions. Nevertheless, the
category “other”, like in French, offers a great diversity of characters and
functions characteristic of the genre: the horse | dragon | beast | demon |
warrior rider | king | priest is.
The science fiction corpus shows other similarities between French
and English. Indeed, in both languages, we find many similar RLTs in
the “action:other” category that are linked to acts involving technolog-
ical objects, for example appuyer | presser | pianoter (sur) le bouton | les
touches | le clavier | la détente (press | touch (on) the button | the keys |
210
V. Goossens et al.
the keyboard | the trigger) and <hit the button>, <pull the plug>, to look
| stare | point at the screen; couper | mettre le contact | moteur (turn off |
switch on the ignition | engine). We also find equivalent motifs for both
languages in the “event” category, such as l’écran | la lampe | le voyant
| la lumière s’allume | s’éteint (the screen | the lamp | the indicator | the
light comes on | goes off) and the lights came | reflect; <the screen show>
or <l’écran montre> (the screen shows); la porte | portière | le sas s’ouvre |
coulisse (the door | the airlock opens | slides) and <the door slides>.
Unlike in the fantasy corpus, very few references to clothing or social
positions are found in French science fiction and then only in the cate-
gories “action:other”, “state” and “other”, for example: <vêtu d’une com-
binaison> (dressed in a jumpsuit), le president est | fait (the president is
| does), être le directeur | un spécialiste (to be the director | a specialist).
Suffice it to note that these are important functions and indicators of
competence or responsibility. In English, the category “action:other”
does not contain any elements relating to clothing and the category
“state” contains only geographical positions. Moreover, we find only
two examples referring to a social position (<be in charge>, <have the
ability>). Similar to Sect. 3.6, the category “other” is also much bigger
here compared to the other genres regardless of language. Also, to be
found here are RLTs containing the extremely polysemous verbs to be
and to have. This unfortunately does not allow us to make a semantic
classification, but it nevertheless does offer a fairly typical panorama of
science fiction usage, with characters (the crew | government | captain is ),
equipment (camera | pod | computer | building is ), space references (the
moon | stars is/are ) and a set of RLTs relating to space travel (the temper-
ature | pressure | atmosphere | gravity is ).
From this panorama stems the rather stereotypical aspect of fantasy
RLTs, especially in French, which immediately guides our reception
of the reading material towards fantasy medievalism, a phenome-
non well-described in various studies. As the medieval universe is not
the only source of inspiration for fantasy (refer to the other subcat-
egories represented in our corpus in Sect. 2), we should not conclude
that fantasy is more stereotypical or more generically marked than sci-
ence fiction. This is why we focus on generic markers: nothing about
these observations is really novel, but they allow us to emphasize some
211
topical motifs. In the next section, we will show that the proposed
semantic categorization also sets the stage for less predictable results to
emerge.
4.2 Less Expected Similarities and Differences
In our semantic profiles (see Tables 4 and 5) two categories appear that
we did not expect to be significant: “cognition”, which is more frequent
in science fiction, and “communication:verbal”, which is found equally
in both genres in French and is specific to fantasy in English.
4.2.1 Decision-Making in Science Fiction and Fantasy
The “cognition” category is a dimension specific to science fiction,

both in the French corpus (18.5% of RLTs) and in the English corpus
(8.67%).
The first observation regarding this category in science fiction reveals
a specific “profile” which seems to support the opposition of reading
postures—the reader’s “awareness” in science fiction and his “confi-
dence” in fantasy (Besson 2007; Saint-Gelais 1999)—which serve as a
criterion for distinguishing science fiction from fantasy. This is of inter-
est to us because of how
– the RLTs highlight a certain definition of thinking (they can be con-

trasted with the RLTs of the fantasy corpus),
– they emphasize the mobilization of thinking and, more specifically in
the case of science fiction, how thinking is a rational process involv-
ing the construction of logical reasoning.
The RLTs in the French science fiction corpus do not take the noun
pensée (thought) as their nominal pivot. Instead, they focus more on
acts of analyzing things: <il a calculé> (he calculated), <il essaye de com-
prendre> (he’s trying to understand), <il en déduit> (he deduced from
it). Thinking is therefore a tool of rationality that makes it possible to
212
V. Goossens et al.
assume, know and invent: <je suppose que c’est>, <je connais par coeur>,
<elle a inventé>.
The RLTs most representative of our science fiction corpus highlight
the development of intellectual reasoning: <mettre au point> (to devise),
trouver la solution | le moyen (to find the solution | the way to) and <il
prend conscience> (he becomes aware of something). These various fea-
tures are shared with the English corpus, where we find <to solve prob-
lem>, <jump to conclusion> or <consider the possibilities> to construct a
reasoning or argumentation, as in this example:
(1) Fedotik said, quickly, We have considered the possibility and have a
solution which we hope will meet with your approval. (Tubb Angado,
1984)
It is noteworthy that the French fantasy corpus contains very different

information on all aspects of the representation of thinking. Very much
akin to imagination and dreams, it is a territory in which one can travel
and even lose oneself: plongé | perdu dans ses pensées (immersed | lost in
his thoughts), <le tirer de ses pensées> (pull him out of his thoughts),
<il reprend ses esprits> (he comes to his senses) or reporter | attirer son
attention (report | draw his attention). The time and memory dimension
also confirms this interpretation, since it concerns <imaginer le pire> (to
imagine the worst), <prendre le temps de réfléchir> (take time to think)
and <elle garde souvenir> (to keep a memory).
The English corpus of science fiction, while more or less sharing the
traits of the French corpus, differs by also pointing to the intellectual
results of reasoning with, for example, <he is convinced> or <make a
mistake>.
As demonstrated by our extracted RLTs, fantasy leaves room for
instinct, whereas science fiction relies on the construction of knowl-
edge through intentionally factual clues. It should also be noted that
awareness in French science fiction is pervaded by essentially dysphoric
feelings or thoughts. It does not highlight hope or joyful discovery, but
rather relates the character to a (sad) reality of deficiencies, gaps and
errors that are sometimes difficult to admit, hence the temporal leap
underlined by the motif.
213
The absence of the “cognition” category from the fantasy corpus

raises questions about how the decision-making process works in these
novels. The observation of the other categories of RLTs recurrent in fan-
tasy possibly furnishes an answer: decision-making is linked to actions
leading to a perception. That is, the characters take in perceptual clues
to interpret and understand a situation and choose a path to follow. For
example, in the French fantasy, the motif humer l’air (sniff the air) takes
place before someone decides which road to take or which stance to
adopt (hostile or friendly), as shown in example (2).
(2) La reine se remit en marche, consciente, cette fois, guettant les bruits,
humant l’air tel un chien de chasse, s’orientant au soleil ou à la mousse
des arbres pour sortir de Brocéliande. (Fetjaine La trilogie des elfes 2 La
nuit des elfes, 1999) (The queen set off again, wide awake this time, alert
to noises, sniffing the air like a hunting dog, orienting herself by the sun
or the moss on the trees to get out of Brocéliande.)
Finally, a set of fantasy RLTs mixing action and perception clearly

stands out in the category “action:other” in French, with <humer l’air>
and inspirer profondément | prendre une (profonde | grande) inspiration
(inhale deeply | take a (deep) breath), <jeter un regard> (take a look),
<observer la scène> (observe the scene), scruter l’horizon | les alentours
(scan the horizon | the surroundings). Similar examples occur in the
English fantasy corpus with have | give a look, look | stare for a moment
| a bit, <look at the clouds> and <stare into the fire>. Hence, we can say
that the analysis of this category is quite productive and rich in generic
markers.
4.2.2 “Communication” in Science Fiction and Fantasy
RLTs we found in the “communication” category generated a number of

unanticipated observations. In French, the “communication” category is
equally represented in both genres but focuses on two different ways of
communicating in them.
214
V. Goossens et al.
Physical communication is sparsely represented, especially in science

fiction, but raises some interesting points of difference. In science fic-
tion the most specific RLTs are <je secoue la tête> (shaking the head) and
<hausser les épaules> (shrugging the shoulders) that are communicative
movements not necessarily directed at anyone and not involving phys-
ical contact. This RLT type is much more diverse in fantasy. We found
communication without physical contact but directed to another per-
son: adresser | jeter | échanger | accorder un regard | signe de tête (address |
throw | exchange | give a look | nod), <faire une reverence> (do a curtsy),
et cetera. We also discovered a good number of RLTs implying physical
contact: <il l’embrasse> (he kisses him), <déposer un baiser (sur le front)>
(kissing (on the forehead)), <caresser sa joue> (caress his cheek), <poser
main sur l’épaule> (put a hand on a shoulder), et cetera.
Verbal communication, by contrast, is a prominent category in both
subgenres. Examining the patterns in this category reveals very differ-
ent modes of communication: in science fiction we find patterns around
distance communication—laisser | recevoir un message (leave | receive a
message), établir le contact | couper la communication (establish contact
| disconnect the call). We also noticed many generic verbs or interac-
tion-structuring verbs: dire d’une voix | sur un ton (say in a voice | in a
tone of voice), <je veux parler> (I want to talk about), <poser la question>
(ask the question), mettre au courant | signaler | faire allusion | faire la
remarque (inform | report | refer to | make the remark). Fantasy offers
us a much more diversified panorama, especially in terms of sound vol-
ume. Naturally, we have motifs built around neutral communication
verbs: dire, parler, demander, prendre | reprendre la parole, et cetera (say,
speak, ask, begin to speak (again)). We also have RLTs involving moral
commitment: je (te le) promets | j’en fais le serment | la promesse, je (dois)
avouer (I promise you, I swear to you, make the promise, I must con-
fess). Finally, we can note a whole range of speaking modalities:
– Silences: garder | imposer le silence (to remain silent), le silence tombe

| s’installe (the silence falls | settles in), <marquer un temps d’arrêt>
(mark a pause time),
– Murmurs: <des murmures s’élèvent> (murmurs rise),
215
– Shouts: pousser un cri de guerre | rage | douleur | un hennissement |

rugissement | grognement, hurler de terreur | rage (shout a war cry | rage
| pain | neighing | roar | growl, scream with terror | rage), étouffer un
cri (mute a scream), et cetera.
This over-representation of the affects and attitudes of FY characters,

as illustrated in the following example 3, is consistent with the results
presented for French in Chapter 4 by Diwersy et al. in this volume for
the study of direct speech verbs in three subgenres including fantasy (as
opposed to romance and crime novels).
(3) Il poussa d’abord un cri de surprise, puis de dégoût, et lança les rep-
tiles grouillants sur le sol en cherchant le coupable des yeux. (Robillard
Les Chevaliers d’émeraude 3. Piège au Royaume des ombres, 2003) (First he
shouted with surprise, then with disgust, and threw the swarming reptiles
on the ground, looking for the culprit.)
In English, the landscape of RLTs denoting verbal communication

(physical communication is almost completely absent) differs on two
levels from the French. First, the “communication:verbal” category con-
tains RLTs only for fantasy, almost none specific to science fiction, like
<that I’m telling the truth> (alongside “communication:undefined” RLTs
such as <haven’t made contact> and able | try to communicate ). Before
attempting to explain this phenomenon, we must first look at the RLTs
specific to fantasy. Unlike in the French corpus, in English 17 of the
30 RLTs annotated as indicating verbal communication are built around
the verb to say: the boy | girl | old man, et cetera says, to say the words,
to say (in) a voice. Furthermore, the other RLTs do not reveal semanti-
cally homogeneous subgroups as is the case in French, even if there are
these similarities: moral commitment (<swore an oath> and <beg your
pardon>) and speaking modalities (<clear his throat>, <give a cry>, <lower
his voice>).
These two findings once again are consistent with the results pre-
sented in Chapter 4 by Diwersy et al. in this volume. Where, on the one
hand, fantasy in both French and English seems to differ from other
216
V. Goossens et al.
genres in terms of communication verbs, English, and in particular

English fantasy, on the other hand diverges from French by overusing
generic verbs and especially the verb to say. This preference for the verb
to say is a possible explanation for the absence of RLTs denoting verbal
communication in science fiction.8 These initial results concerning both
convergences and divergences with regard to communication in the fan-
tasy genre in the French and English-speaking traditions merit being
developed further in the rest of the project.
Having completed our analysis of fantasy and science fiction using
semantic categorization, we will now turn to the methodological impli-
cations of our approach.
5 Methodological Outlook
The semantic grid we presented in this article, while admittedly suffer-
ing from some limitations, is nevertheless an effective tool for analyz-
ing fiction. Negotiated within a multidisciplinary team of researchers, it
raises the following issues:
– Methodological: how to put into perspective and apply achievements

and concepts of corpus semantics to the statistical extraction of RLTs?
– Practical: how to describe the results and problems specific to each
pilot study which used the semantic grid?
– Interpretative: how to find interesting and relevant items for the
extended stylistic study of the selected corpus?
The complexity of the semantic classification is also driven by that of

the PhraseoRom project itself, whereby the latter applies the statistical
research methodology of extracting lexico-grammatically complex RLTs
8Since French is characterized by a great lexical diversity around the verbs of speaking, we can
observe a distribution of RLTs in the different genres. The fact that the English novel gives prom-
inence to the verb to say means that, if we compare two corpora as in this article, the RLTs built
around this verb, although found in both sets of novels, are considered specific to fantasy because
of a greater number of occurrences.
217
in studying the heterogeneous corpora of multilingual narrative fiction

(see Chapter 2 by Legallois and Koch in this volume).
The complexity of the semantic grid in turn raises a number of other
questions:
1. Should we use and/or adapt existing semantic grids, singly or in com-

bination, that were proven and tested on other projects? We were
unable to find grids that we could use “as is” for our project, which
forced us to develop a methodology that would integrate pre-exist-
ing semantic typologies (for example Mahlberg 2007), but, above all,
would let us devise our own semantic categories. The main issue was
that the categories we selected, even if they did not per se present
an impeccable semantic typology, had to be relevant and effective for
working on our specific PhraseoRom corpus while also reflecting its
inherently complex multilingual and contrasting dimension.
2. How should the semantic grid allow for and integrate the results of
the pilot studies that it was used for? Our answer here is that the
grid will benefit from the results of studies which use it and will be
adapted accordingly.
3. How and to what extent should the semantic grid deal with the dis-
cursive specificity of the texts in the PhraseoRom corpus, which is
composed of several subgenres that differ greatly from contemporary
narrative fiction. For example, should we create a semantic subcate-
gory based on the analysis of a specific subgenre or not?
4. How to analyze sometimes incomplete structures (for example
<qu’elle porte> (that she wears/carries/holds), which can correspond
to sequences as varied as l’enfant qu’elle porte (the child she is car-
rying)/qu’elle porte la main à son coeur (that she holds her hand to
her heart)/qu’elle n’a jamais porté de chaussures (that she’s never worn
shoes before), et cetera) or, on the other hand, very rich structures
(for example <ajoute-t-elle avec un sourire> (she adds with a smile))
that combine verbal and physical communication and facial move-
ment? Or, for that matter, how should we approach structures that
combine polysemy and complex meanings (for example <il se remet>,
which can refer either to an action that combines movement with
218
V. Goossens et al.
another type of action (he starts again) or a qualia that combines sen-
sation and affect (he recovers))? Indeed, our methodology for extract-
ing RLTs implies giving consideration to both the lexical basis of the
structures and a more abstract morpho-syntactic framework for them
(see Chapter 2 by Legallois and Koch in this volume).
The available space for this article precludes our addressing all these
questions in detail here, but we will take up some of them in what fol-
lows. The current version of the semantic grid reflects the experience
gathered in the course of its practical application. It benefited from
the negotiations with and feedback from its users relating to its cate-
gories. Part of the process consisted of holding corpus harmonization
meetings for solving annotation problems specific to each language and
subgenre we studied. The bottom line is that the semantic grid is well
on its way to becoming one of the essential tools for harmonizing the
PhraseoRom project in its multiple contrasting dimensions. The exam-
ple of the “action” category can be instructive in this regard. We divided
it into four different subcategories: “action:travel”, “action:movement”,
“action:other” and “action:undefined”. The distribution into those four
action categories relates to the specificity of the PhraseoRom corpus
and, indeed, dovetails with that project’s inherent interpretative choice
of analyzing works of fiction. As normally happens in any annotation
process (Rastier 2001; Bachimont 2014), the choices we made in ours
may have filtered out a certain number of results yet they also produced
just as many new ones. To summarize, our results are conditioned by
corpus-specific research issues and the above-mentioned steps involved
in negotiating the semantic grid.
To recap, the “action:movement” and “action:travel” subcategories
make it possible to describe in detail actions that seemed characteris-
tic of the subgenres analyzed by the PhraseoRom project. This distri-
bution within the “action” category also holds the potential for making
our annotation system interoperable: to reflect the different semantic
characteristics of such different fiction subgenres as historical novels,
romance, crime fiction, fantasy, science fiction or general literary fiction
without overwriting them.
219
To sum up, the heuristic contribution of our semantic grid confirms

hypotheses created by extant literary analyses. This is particularly evi-
dent in our section on generic markers. Admittedly, some of the results
are expected but, thanks to our extraction methodology, they can be
seen as confirming the reading hypotheses that allow us to distinguish
science fiction and fantasy in that respect. Furthermore, it brings out
other, less expected categories such as “cognition” and “communica-
tion:verbal”. Finally, it allows us to discuss some results (such as the
contents of the categories “action:other” and “event”), to examine them
more closely, and to refer to the meaning of certain RLTs and their
discursive function in context. For instance, it makes the distinction
between English and French as well as the discursive and lexical particu-
larity of fantasy especially interesting to describe.
6 Conclusion
Our analysis of RLTs and motifs in this article has allowed us to distin-
guish the two subgenres science fiction and fantasy that make up the
“littérature de l’imaginaire”. In follow up, we propose to refine these
literary “profiles” by comparing each specific subgenre with the entire
PhraseoRom corpus. This will better highlight their phraseological simi-
larities and question their common points by problematizing them.
Literary critics do not rely on linguistic differences to distinguish
between science fiction and fantasy but instead on an opposition of
the imaginary and of reader experience. Our semantic analyses yielded
the following: they made it possible to highlight what we term generic
markers (the medievalism of fantasy, the salience of specific spaces
(space, forest) and certain so-called stereotypical actions—for instance,
the warlike dimension of fantasy actions). Our analyses also confirmed
the distinction between two “régimes de sens” expressed in the two
distinct categories of cognition in science fiction and the perceptual
dimension of “action:other” in fantasy.
Finally, we observed that the line separating the two genres is gen-
erally respected in both the French and English corpora. However, if
220
V. Goossens et al.
we examine certain categories closely (such as “cognition” and espe-

cially “communication”), we see specificities pertaining to each language
emerge between the English and the French corpus.
Going forward, we plan to study the PhraseoRom’s remaining genres
using the methodology developed for the present analysis. After com-
pleting this project, we will be in a position to determine if our seman-
tic grid is a suitable tool for surfacing distinctions between other genres
as relevant as those we were able to draw between science fiction and
fantasy.
References
Adam, Jean-Michel. 2005. Les Textes: types et prototypes. Récit, description, argu-
mentation, explication et dialogue. Paris: Armand Colin.
Bachimont, Bruno. 2014. “Le Nominalisme et la culture: questions posées
par les enjeux du numérique.” In Digital Studies: Organologie des savoirs et
technologies de la connaissance, edited by Bernard Stiegler, 63–78. Paris: FYP
Editions.
Baroni, Raphaël. 2003. “Genres littéraires et orientation de la lecture: Une
lecture modèle de ‘La mort et la boussole’ de J.L. Borges.” Poétique 134:
141–57.
Baudou, Jacques. 2005. La Fantasy. Paris: PUF.
Bazin, Laurent. 2015. “Pluralité des mondes, porosité des genres: poétique du
possible dans les littératures contemporaines de l’imaginaire.” In Poétiques
du merveilleux: Fantastique, science-fiction, fantasy et littérature et dans les arts
visuels, edited by Anne Besson and Évelyne Jacquelin, 107–20. Arras: Artois
Presses Université.
Besson, Anne. 2007. La Fantasy: 50 Questions. Paris: Klincksieck.
Cornillon, Claire. 2012. Par-delà l’infini: La Spiritualité dans la science-
fiction française, anglaise et américaine. PhD diss., Université de la Sorbonne
Nouvelle, Paris III.
Eco, Umberto. 1979. Lector in fabula. Paris: Livre de Poche.
Sorba. 2018. “Motifs textuels spécifiques au genre policier et à la littéra-
ture «blanche».” In 6ème Congrès Mondial de Linguistique Française, edited
by Franck Neveu, Bernard Harmegnies, Linda Hriba, and Sophie Prévost.
221
SHS Web of Conferences 46: https://www.shs-conferences.org/articles/

shsconf/pdf/2018/07/shsconf_cmlf2018_06007.pdf.
Goossens, Vannina, Clémence Jacquot, and Iva Novakova. 2018. “Distinguer
les genres de la littérature de l’imaginaire par les motifs textuels.” Journée
d’étude Lexique et frontières de genres. Pau, 22 March 2018.
Hommel, Élodie. 2017. Lectures de science-fiction et fantasy: enquête sociologique
sur les réceptions et appropriations des littératures de l’imaginaire. PhD diss.,
Université de Lyon.
Mahlberg, Michaela. 2007. “Corpus Stylistics: Bridging the Gap between
Linguistics and Literary Studies.” In Text, Discourse and Corpora: Theory and
Analysis, edited by Michael Hoey, Michaela Mahlberg, Michael Stubbs, and
Wolfgang Teubert, 219–47. London: Continuum.
Oxford Living Dictionaries English. “Cognition.” Oxford: Oxford University
Press. https://en.oxforddictionaries.com/definition/cognition (Oxford dic-
tionary online).
Rastier, François. 2001. Arts et science du texte. Paris: PUF.
Saint-Gelais, Richard. 1999. L’Empire du pseudo. Modernités de la science-
fiction. Québec: Nota Bene.
Torres, Anita. 1997. La science-fiction française, Auteurs et amateurs d’un genre
littéraire, Logiques sociales. Paris: L’Harmattan.
8
Reading and Writing as Motifs
in English and French General Fiction
Julie Sorba, Laetitia Gonon, Susanne Dyka
and Vannina Goossens
1 Introduction
To study the phraseology of literary fiction in this chapter we use the
crime novels corpus as a contrast corpus. Previously, Gonon et al.
(2018) identified specific phraseological units for literary fiction (GEN)
as opposed to those of crime novels (CRIM) in French. These units take
the form of Lexico-Syntactic Constructions (LSC) with the core verb
J. Sorba (*) · L. Gonon · V. Goossens

e-mail: julie.sorba@univ-grenoble-alpes.fr
L. Gonon
V. Goossens
S. Dyka
https://doi.org/10.1007/978-3-030-23744-8_8
224
J. Sorba et al.
écrire: écrire_DET_lettre (write_DET_letter ) and écrire_DET_roman

(write_DET_novel ). Their study revealed that these constructions imple-
ment specific motifs in literary novels. The way authors depict fictional
(roman ) or intimate (lettre ) writing activities in them clearly shows the
attention they pay to how the language works.
Building on those previous results, in this paper we aim to compare
the LSCs centred on the French verbs lire and écrire to similar con-
structions around read and write in English novels. This will let us ana-
lyze the role they play in the structuring of French and English novels.
We begin by presenting first the theoretical framework of our study
(Sect. 2), then the corpora and the text mining methodology we applied
to them (Sect. 3) and, finally, our results (Sect. 4).
2 Theoretical Framework
Since Sinclair’s pioneering approach (1991) to studying phraseological
units, the field has evolved from dealing with “lexical sequences per-
ceived as preconstructed” (Legallois and Tutin 2013, 3, our translation)
to one expanded with diversified objects of study:
La phraséologie intègre désormais des objets d’étude très variés, allant

des collocations aux séquences discursives en passant par la parémiologie,
ou encore, les schémas syntaxiques. Les approches proposées s’étendent
maintenant au-delà des disciplines traditionnelles de la lexicologie, de
la syntaxe et de la sémantique, et abordent largement la linguistique du
discours, la psycholinguistique ou la linguistique informatique (Legallois
and Tutin 2013, 3). (Phraseology now includes a wide range of study
objects, from collocations to discursive sequences, paremiology and syn-
tactic patterns. The proposed approaches now extend beyond the tradi-
tional disciplines of lexicology, syntax and semantics, and broadly address
the linguistics of discourse, psycholinguistics or computer linguistics, our
translation.)
This diversification has been accompanied by an exponential growth

in the terminology as recently surveyed by Zaharieva and Kaldieva-
Zaharieva (2017). We adopt here for our purpose an expansive
8 Reading and Writing as Motifs in English and French General …
225
approach to the phraseology that Novakova and Sorba (2018, 55) in

their study on the emotional lexicon showed was relevant for analyz-
ing discursive phenomena: “Le sens des lexies d’affect en discours peut
être envisagé comme un phénomène continu opérant sur différents
niveaux qui interagissent entre eux et qui mettent en jeu des unités de
rangs différents (syntagme, phrase, paragraphe, texte).” (The mean-
ing of the emotional lexicon in discourse can be seen as a continuous
phenomenon operating on different levels that interact with each other
and involve units of different levels (phrase, sentence, paragraph, text),
our translation). Therefore, since we are interested in the functioning
of phraseological units at both the microstructural (phrase, sentence)
and macrostructural (paragraph, text) levels, we adopt a global approach
inspired by British functionalist and contextualist models (Sinclair
2004; Hoey 2005; Mahlberg 2007). At the microstructural level, we
consider phraseology, as Grossmann et al. (2017, 7) do, as “la congru-
ence à la fois syntaxique et sémantique qui lie les unités lexicales entre
elles pour donner lieu à des unités polylexicales qui se distinguent par
une fixité d’emploi conditionnant leur fonctionnement interne et leur
combinatoire externe.” (The syntactic and semantic congruence that
binds lexical units together to give rise to polylexical units that are dis-
tinguished by a fixed use conditioning their internal functioning and
external combinatory, our translation.) Consequently, we propose as the
first step in our analysis a detailed study of the lexico-syntactic combi-
nations of the different units in both corpora. Following Blumenthal,
we consider the combinatory of a lexeme to be its statistically signifi-
cant association with specific collocates (Blumenthal 2006, 15). This
will allow us to determine how the LSCs operate phraseologically and
how each of them “prime to occur” with particular other lexical units
(Hoey 2005). Our goal is to discover whether these LSCs are part of an
extended phraseological unit that we call motif after Longrée and Mellet
(2013, 66):
[L]a notion de motif est conçue comme un moyen de conceptualiser la

multidimensionnalité (ou le caractère multi-niveau) de certaines formes
récurrentes qui sollicitent à la fois le lexique, les catégories grammaticales
et la syntaxe, éventuellement la prosodie, la métrique. … un motif admet
226
J. Sorba et al.
toutefois divers types de variations et sa micro-structure combine à la fois

des éléments de stabilité assurant sa mémorisation et sa reconnaissance,
et des éléments de transformation assurant le jeu inhérent aux divers
usages en discours. … Sur le plan fonctionnel, le motif est un “cadre col-
locationnel” accueillant un ensemble d’éléments fixes et de variables, sus-
ceptible d’accompagner la structuration textuelle et, simultanément, de
caractériser des textes de genres divers, voire de permettre la détection,
au sein d’un même texte, des passages de registres différents. (The notion
of motif is conceived as a way of conceptualizing the multidimensional-
ity (or multi-level aspect) of certain recurring forms that involve the lexi-
con, grammatical categories and syntax, possibly prosody, and metric. …
a motif, however, admits different types of variations and its microstruc-
ture combines both elements of stability, ensuring its memorization and
recognition, and elements of transformation ensuring its various uses in
discourse. … Functionally, the motif is a “collocational framework” con-
taining a set of fixed elements and variables, capable of accompanying
textual structuring and, at the same time, characterizing texts of various
genres or even allowing the detection, within the same text, of passages
from different registers, our translation.)
Next, our analysis highlights the paradigmatic and syntagmatic varia-

tions of LSCs to establish the first criterion for identification of motifs.
Once this has been accomplished, we propose to establish the second
criterion for identification of motifs by analyzing the discursive func-
tions of these units. Indeed, as a phraseological phenomenon, the motif
is of crucial importance in the complexity of linguistic arrangements
(Legallois and Tutin 2013, 9). The discursive function of a unit depends
on its role in relation to the cotext: it maintains thematic coherence
and continuity in a text (Martin 1983, 15–16). In our study, we under-
stand discursive function as relating to fiction discourse (see Appendix
B). Indeed, in the novel, discursive functions are above all narrative and
descriptive (Adam 2011, 267): “[U]n texte à dominante narrative est
généralement composé de relations d’actions, d’événements, de paroles
et de pensées représentées, mais … il comporte aussi des moments
descriptifs plus ou moins développés” (a predominantly narrative text
is generally composed of related actions, events, words and thoughts
represented, but … it also includes more or less developed descriptive
227
moments, our translation). That said, motifs can have other discursive
functions in the novel: for example, the so-called commentary function1
in which the motif is part of a reflective statement about the writing
practised by a character, or the so-called cognitive function in which
the motif is found in a cotext reflecting the thoughts, questions and
hypotheses formulated by a character (Gonon et al. 2018; Gonon and
Sorba, forthcoming).
With the data collected during these three stages, we can proceed
with characterizing how these phraseological units operate in the liter-
ary fiction corpora. This will enable us to contribute to “an operating
theory of genres,” in Rastier’s words (2011, 72, our translation), since
the fact of a text belonging to a genre conditions the lexical, morpho-
syntactic and discursive variations found as compared with other gen-
res (see, for example, Biber 1993; Malrieu and Rastier 2001; Siepmann
2015, 2016).
3 Corpus and Methodology

The French and English fiction corpora on which we base our study are
presented in detail in Appendix A. To explore these corpora, we used
the Lexicoscope2 text mining tool for annotated corpora in syntac-
tic dependencies (Kraif 2016). This interface uses an iterative method
for extracting Recurrent Lexico-Syntactic trees (RLTs) whose relevance
for detecting significant statistical associations linking several words by
a syntactic relationship was demonstrated by Tutin and Kraif (2016).
It calculates the specificity of associations on the basis of the statistical
index log-likelihood ratio (LLR), whose utility for text mining Dunning
(1993) has shown.3
We used the Lexicoscope to contrast the GEN and CRIM corpora
by extracting RLTs filtered for a length greater than 3 words, dispersion
1An exhaustive list of the discursive functions with their definition can be found in Appendix B.
2http://phraseotext.univ-grenoble-alpes.fr/lexicoscope/.
3On this methodology, see also Chapter 2 by Legallois and Koch in this volume.
228
J. Sorba et al.
Table 1 Specificities of the recurrent lexico-syntactic patterns of our study in

the GEN corpora
RLT f CRIM f GEN Disp CRIM Disp GEN LLR
<lire_le_lettre> 20 98 13 51 17.5864
<écrire_un_roman> 26 123 11 42 20.8373
<je_écrire_lettre> 13 83 10 43 21.2678
<je_lire_livre> 13 94 7 45 27.4631
<read_the_letter> 23 77 16 35 18.7635
<write_the_word> 4 33 3 14 19.2308
<read_the_book> 11 52 7 30 19.6612
<write_the_letter> 3 38 3 19 27.2125
greater than 3 novels, and frequency greater than 5 occurrences. This

first step yielded a total of 32,699 such patterns for English and 71,635
for French.4
To target specific patterns we next applied the following filters to the
French corpora in a previous study (Gonon et al. 2018):
(i) a 10.83 LLR threshold, above which the association cannot be con-
sidered as random;
(ii) a dispersion criterion to keep only the patterns present in at least
50% of the CRIM texts and 25% of the GEN texts;
(iii) a grammatical criterion for the mandatory presence of a verb to
exclude collocations already identified as commonly found ones (for
instance, scène de crime in the CRIM).
This second stage produced 264 patterns in the French GEN corpus. In
this list, we observed several patterns organized around the pivots lire
and écrire. We therefore opted for studying them in detail and contrast-
ing them with the patterns of the English corpora. We applied identical
criteria of specificity and presence of a verb for GEN English corpora to
search for patterns with the verbs read and write. The specificity of each
of these patterns is presented in Table 1.5
4To explain this significant difference between the results, see Chapter 7 by Goossens et al. in this
volume.
5Legend: f = frequency of the recurrent lexico-syntactic pattern (number of occurrences);
disp = dispersion of the recurrent lexico-syntactic patterns (number of authors).

229
In the next section, we begin the presentation of our results by ana-

lyzing the potential lexico-syntactic combinations of the constructions
containing lire/read and écrire/write to show how these phraseological
units are employed in discourse. We then proceed to analyze the syn-
tagmatic and paradigmatic variations of the LSC to establish their motif
status. Lastly, we examine the discursive functions of these motifs.
4 Results
4.1 Study of Lexico-Syntactic Constructions
and Variations
To reiterate, we studied the potential lexico-syntactic combinations of

the RLTs to see how they are implemented in discourse to acquire the
status of LSCs. Table 2 lists the statistically significant collocates we
extracted with Lexicoscope for each construction (in brackets, the LLR;
the last column shows the total collocates count).
Lexicoscope revealed that the specific combinations with grammatical
collocates exceed those with lexical collocates in both the French and
English GEN corpora. For LSCs with a verbal core, as in this case, spe-
cific combinations with subject pronouns (il, je, tu, vous, he ) or object
pronouns (lui, vous ) and with auxiliaries allowing conjugation (avoir,
have ), are quite predictable.
Even so, these results already indicate some initial trends. In the
French corpus, we notice in fact that when realized in discourse écrire_
un_roman prefers the first person (LLR 90.5592):
(1) Moi aussi j’ai écrit un roman: mon unique roman, rédigé à l’âge de
neuf ans, un an après la mort de mon père, dans un cahier d’écolier à cou-
verture rouge. (Millet La Fiancée libanaise, 2011) (I too wrote a novel: my
only novel, composed at the age of nine, a year after my father’s death, in
a school notebook with a red cover.6)
230
J. Sorba et al.
Table 2 The lexical and grammatical collocates of the eight patterns

Lexical collocates Grammatical collocates Total
French
<lire_le_lettre> 0 avoir (30.6636), il (38.3204), 4
je (20.3531), pas (10.8956)
<écrire_un_roman> devoir (13.9155) je (90.5592), vous (30.7597), 6
tu (15.5508), de (12.3343),
que (11.3022)
<je_écrire_lettre> long (54.6962), un (181.6025), ce (54.5553), 6
pouvoir (20.7878) lui (50.0148), vous (16.1732)
<je_lire_livre> vouloir (20.2510), avoir (100.5728), un 5
savoir (14.6914) (44.3225), tout le (33.0541)
English
<read_the_letter> again (32.3000), through (34.9124), when 8
begin (25.7419), (21.0263), he (12.0410), that
write (19.2003) (11.5422), and (11.4589)
<write_the_word> 0 which (26.3315) 1
<read_the_book> 0 all (32.2536), have (29.1302) 2
<write_the_letter> 0 0 0
This preference for the first person in the French corpus is striking: not
only do the two patterns <je_écrire_lettre> and <je_lire_livre> from the
extraction phase display this preference but so does the Lexicoscope,
listing the French pronoun je among the most specific collocates of the
other two LSCs (LLR 90.5592 for écrire_un_roman and LLR 20.3531
for lire_le_lettre ).
In addition, the presence of the collocate avoir reflects the frequent
use of compound tenses, marking a completed aspect for the two LSCs
containing the French verb lire (2), which is not the case for the two
LSCs containing the French verb écrire. Reading is often the trigger for
another action (in the narrative sequence), and the completed aspect of
the verb shows its anteriority in relation to subsequent actions.
(2) Elle avait lu la lettre et l’avait fourrée dans sa poche. (Modiano L’Herbe
des nuits, 2012) (She had read the letter and stuffed it in her pocket.)
In the English GEN corpus, these initial results testify to a specific link
between the LSCs read_the_letter and write_the_word and the syntactic
231
subordination structure (when LLR 21.0263, that LLR 11.5422, which

LLR 26.3315).
(3) When Omar Khayyam read the letter, his first reaction was to whistle
softly with something very like admiration. (Rushdie Shame, 1983).
(4) A sheet of blank notepaper on which were written the words.
(Woodward I’ll Go to Bed at Noon, 2004)
Concerning lexical collocates, three of the four LSCs in the French cor-
pus come with modal verbs (devoir, pouvoir, vouloir ).
(5) Je voudrais bien lire les livres que vous me donnez, et la nuit, dès que
vous êtes parti, elle éteint la lampe et se met à parler, interminablement,
sur les uns et les autres… (Déon Un taxi mauve, 1973) (I would like to
read the books you give me, and at night, as soon as you leave, she turns off
the lamp and starts talking, endlessly, about everyone…)
The example (5) links this observation to the one previously made
about the privileged use of the first person. It would seem that the LSCs
including the verbs lire and écrire in the French GEN corpus strongly
involve the enunciator in his or her discourse.
4.2 From Lexico-Syntactic Constructions to Motif:

Syntagmatic and Paradigmatic Variations
Our analysis of syntagmatic and paradigmatic variations affecting LSCs

shows that they meet the first condition that qualifies them as potential
motifs.
4.2.1 The Syntagmatic Variations
Within LSCs, the syntagmatic variations can affect the verbal form
as well as the nominal form and its determiner. The data presented in
Table 3 show that the determiner varies across the eight LSCs.
232
J. Sorba et al.
Table 3 Syntagmatic variations of the article across the eight LSCs

Definite article Indefinite article Other article Total
French
lire_le_lettre 92 (72%) 12 23 127
écrire_un_roman 10 123 (89%) 4 137
je_écrire_lettre 0 62 (75%) 20 82
je_lire_livre 34 (36%) 30 30 94
English
read_the_letter 60 (55%) 3 45 108
write_the_word 33 (91%) 3 0 36
read_the_book 44 50 175 (79%) 219
write_the_letter 39 145 (78%) 0 184
As it turns out, in the two French LSCs, the verb lire favours the defi-
nite article (la lettre 72%/le livre 36%) while the verb écrire favours the
indefinite article (une lettre 75%/un roman 89%). On the other hand, in
the English corpus, the verb to write does not express a clear preference
for associating either the indefinite article (a letter 78%) or the definite
article (the word 91%). In the English GEN corpus we cannot find a
clear preference for the definite article for read_the_letter and for read_
the_book. Both patterns occur with other articles; with read_the_book
we found 52 cases without article, which, of course, is a consequence of
the English plural.
Table 4 summarizes the variations of the grammatical number of
nouns.
In the French corpus, the noun lettre favours the singular form with
the two verbs lire (66%) and écrire (75%) as does roman with écrire
(78%). By contrast, in the LSC je_lire_livre, the noun appears mainly
in the plural form (61%), a difference that requires further investigation
to be interpreted. In the English corpus, the noun after the verb write
favours the singular form (56% for word and 97% for letter ), whereas
letter occurs more often in singular with read than in plural and book
roughly does so evenly in singular and plural.
The third variable element is the verbal core. As shown in Table 5, the
verbal core of each LSC can be found either at a conjugated tense or at
the infinitive or participle.
233
Table 4 The syntagmatic variations of the noun across the eight LSCs
Singular Plural Total
French
lire_le_lettre 84 (66%) 43 127
écrire_un_roman 107 (78%) 30 137
je_écrire_lettre 62 (75%) 20 82
je_lire_livre 36 58 (61%) 94
English
read_the_letter 83 (76%) 25 108
write_the_word 20 (56%) 16 36
read_the_book 107 112 (51%) 219
write_the_letter 179 (97%) 5 184
Table 5 The syntagmatic variations of the verb across the eight LSCs
Tense Infinitive Participle Total
French
lire_le_lettre 77 (61%) 41 9 127
écrire_un_roman 58 79 (57%) 0 137
je_écrire_lettre 65 (79%) 17 0 82
je_lire_livre 86 (91%) 8 0 94
English
read_the_letter 89 (82%) 18 1 108
write_the_word 31 (86%) 3 2 36
read_the_book 170 (77%) 47 2 219
write_the_letter 131 (71%) 29 24 184
The verb in the LSC is mostly conjugated—most often in a predict-

able way—to a past tense, which is the tense par excellence of the nar-
rative. Only the LSC écrire_un_roman favours the infinitive form in its
occurrences. For the English patterns, the verbs occur more often in one
of the tenses, very often in the past tense, which is the narrative tense
in English as it is in French. The infinitive is also found quite often
with read and write with the exception of write_the_word where cases of
infinitive use are scarce.
The syntagmatic study of the eight distributions shows that their con-
stituent elements do vary but that this variation does not apply in the
same proportions.
234
J. Sorba et al.
4.2.2 The Paradigmatic Variations
The paradigmatic variations affect the core of the LSCs; it can be:
– a modifier of the noun (see example 6, spirited nonchalant letters, which

I have such difficulty in answering; and elsewhere in the English corpus
love letter, long ~, brief ~, formal ~, fierce ~, stilted ~, suitable ~, similar ~,
business like ~, careful ~, finely argued ~, big ~, hysterical ~ for instance;
and in the French corpus lettre d’amour (love letter) ~ qui m’a coûté des
jours et des nuits (that took me days and nights), ~ de sa fiancée (from
his fiancée), ~ d’Algérie (from Algeria); nouveau roman (other novel),
roman dans lequel les révolutionnaires mexicains pourchassent et fusil-
lent les curés (novel in which Mexican revolutionaries chase and shoot
priests), roman à l’eau de rose (cheesy novel) for instance), or
– a circumstant (see example 7, souvent (often) and elsewhere in the
English corpus never write a letter, simply ~ for instance; and in the
French corpus lire une lettre fort attentivement (read a letter very care-
fully), ~ à mi-voix (mid-voice); écrire un roman en cachette (write a
novel in secret), ~ en deux jours (~in two days) for instance).
(6) Melissa still writes the spirited nonchalant letters which I have such diffi-
culty in answering save by whining retorts about my circumstances or my
improvidence. (Durrell The Alexandria Quartet, 1962)
(7) M. Bentley avait un vrai talent d’écriture et le révérend nous lisait sou-
vent ses lettres avant que nous ne priions pour lui. (Autissier L’Amant de
Patagonie, 2012) (Mr. Bentley had a real talent for writing and the rever-
end often read his letters to us before we prayed for him.)
Table 6 summarizes how much the variation for each LSC affects the
right and left contexts of the noun:
Mostly, it is the right context of the noun that varies. It is then neces-
sary to specify:
– the circumstances of the act of writing or reading:

(8) George wrote the letter rapidly, straight out, in a state of excitement as if
inspired. (Murdoch The Philosopher’s Pupil, 1983)
235
Table 6 The paradigmatic variations across the noun in the eight LSCs
Left context Right context Bare Total
French
lire_le_lettre 11 76 (60%) 40 127
écrire_un_roman 22 53 62 (45%) 137
je_écrire_lettre 19 63 (77%) 6 82
je_lire_livre 12 48 (51%) 34 94
English
read_the_letter 10 89 9 108
write_the_word 2 27 (75%) 8 36
read_the_book 26 165 109 219
write_the_letter 29 133 (72%) 22 184
– the properties of the letter, word or book (source, origin, content,

author),
– an axiological judgement:
(9) Il écrivait un roman très compliqué: c’était délicieux. (Defalvard Du

temps qu’on existait, 2011) (He was writing a very complicated novel: it was
delightful.)
Left context for the English patterns reveals mostly adjectives and some
nouns in the case of read_the_book as premodification of letter and book.
All of our observations on syntagmatic and paradigmatic variations
confirm that these eight LSCs fulfil the first criterion for qualifying as
motifs. We can now analyze the discursive functions of all eight by con-
trasting the two corpora.
4.3 Discursive Functions of the Motifs
The discursive functions (DFs) of the studied motifs are presented

in a way that allows comparing the EN corpus and the FR corpus.
Comparing the discursive functions across languages can help deter-
mine whether the motifs are common to the literary fiction or whether
they instead represent a specificity of French novels as opposed to
English ones, and vice versa.
236
J. Sorba et al.
The discursive functions most represented in the novel genre are nar-
rative and descriptive: they describe places, time periods, and charac-
ters whose story is told. However, in CRIM, the plot is dominated by
investigation and therefore by action. By contrast, the specific patterns
of GEN FR show that these novels of French literature often stage a
reflection about themselves, with a statistically significant metafictional
dimension (Gonon et al. 2018): the discursive functions of the motifs
are less often narrative or descriptive, even if the latter remain highly
represented. In the following sections, we analyze the variety of DFs for
the motifs in the two GEN corpora.
4.3.1 Reading as Motif
Lire_DET_lettre and read_DET_letter

The discursive function of the motif lire_DET_lettre in GEN FR is
often narrative in nature:
(10) Quand Véro est accourue, ce sale matin de janvier 17, venant d’ap-
prendre par quelqu’un du quartier que son amant était mort, Petit Louis
lui a fait lire la dernière lettre de l’Eskimo [où il exprime ses dernières
volontés] et demandé de s’expliquer. Elle était en larmes, à genoux sur
le sol, effondrée. (Japrisot Un long dimanche de fiançailles, 1991) (When
Véro came running, on that bloody morning in January 17, after hearing
from someone in the neighbourhood that her lover had died, Petit Louis
made her read the Eskimo’s last letter [where he expressed his last wishes]
and asked for an explanation. She was in tears, kneeling on the ground,
devastated.)
The motif in its expanded version (lui a fait lire la dernière lettre de l’Es-
kimo: made her read the Eskimo’s last letter) here establishes a coherence
between the different characters through the causative construction.
Petit Louis is the subject, the indirect object lui designates Véronique,
and the epithet dernière and the possessive structure de l’Eskimo are
nominal expansions that make the voice of the departed heard (Kleber,
called l’Eskimo ). The motif therefore has a direct consequence in the
237
narrative sequence: it provokes the reactions of Petit Louis (in the clause
coordinated by et ) and Véronique (in the following sentence). In this
way, it advances the plot, as is often the case with this motif in GEN:
the letter reveals something, which has consequences for the reader. For
the same reason it is also the most represented discursive function in the
English corpus:
(11) She turned over the other letter. It was addressed to herself. As she
instantly recognised the writing, she stood for several moments very still.
Then breathing deeply she moved into her bedroom and closed the door.
She sat down on the bed, opened the envelope, and read the letter through
carefully. (Murdoch The Green Knight, 1993)
The motif in the past tense appears in a list of actions that lead to the
reading of the letter itself, apparently feared by the character, and which
will influence events later in the novel.
However, the image of a character reading a letter also lends itself to
description in GEN FR. For example, in this passage, the narrator reads
a book while observing the woman who shares his room:
(12) Un doigt posé sur la bonne ligne, j’attendais pour continuer ma lec-
ture. En tournant sur elle-même, Edmondsson lisait des lettres, classait des
documents. Elle s’éloignait du bureau, revenait vers moi. Elle s’asseyait sur
le fauteuil et, en bougeant les lèvres, prenait connaissance d’un imprimé;
puis elle décroisait les jambes, se relevait et faisait des commentaires.
(Toussaint La Salle de bain, 1985) (One finger on the right line, I waited
to continue my reading. Turning around, Edmondsson read letters, filed
documents. She walked away from the desk, came back to me. She would
sit in the chair and, moving her lips, read a printout; then she would
straighten her legs, stand up and make comments.)
The protracted reading (by the narrator) of a book is interrupted by

Edmondsson’s agitation: she reads letters while also doing small actions
without any consequence for the plot, but the narrator tracks her move-
ments. The motif therefore has a descriptive discursive function, like the
other clauses in imperfect tenses in which Edmondsson is the subject.
238
J. Sorba et al.
The indefinite plural article des, on the one hand, signifies agitation (she
passes quickly from one letter to another), and, on the other, hints at
the lack of importance she ascribes to this act of reading, since the con-
tent of the letters remains unknown to the reader.
In GEN EN, on the other hand, there are far fewer significant
occurrences of this descriptive discursive function: reading a letter is a
dynamic motif rather than a justification for describing a character.
Admittedly, some uses of this motif are more difficult to identify in
terms of discursive function. In the following example, the narrator has
dinner with a woman he wants to seduce. He apologizes for not being
able to cook by highlighting his cultural knowledge when recounting
his trip to the shops:
(13) Je pense à Diogène qui offrit des figues à Platon, et à La

Rochefoucauld qui envoyait des Maximes inédites à une amie sans pré-
tendre pour autant mériter son potage de carottes. Quand il ne réussissait
pas à trouver les truffes qu’elle lui demandait, il lui envoyait à la place
quelques Maximes qui, s’excusait-il, ne valaient pas de bonnes truffes. Si
nous nous revoyons, je vous lirai les lettres qu’il écrivit à cette femme, la
marquise de Sablé. (Weyergans Trois jours chez ma mère, 2005) (I think of
Diogenes who offered figs to Plato, and La Rochefoucauld who sent origi-
nal Maximes to a lady friend without claiming to deserve her carrot soup.
When he couldn’t find the truffles she asked him for, he sent her some
Maximes instead which, he apologized, were not worth good truffles. If
we meet again, I will read the letters he wrote to this woman, the Marquise
de Sablé, to you.)
The motif of reading letters creates a link between two characters (je
subject, vous object) around the act of reading in the story, but the
discursive function is not narrative: the sentence is indeed hypotheti-
cal, and the future is not preordained. It is subject to how successful
François is in seducing Juliette. It even has no value except as seduction
for its own sake (once his goal is reached, François will probably never
read these letters to Juliette). The motif is therefore a false promise (but
its utterance is not a lie, it is still a promise, see Austin [1970, 45]): the
discursive function of the motif is pragmatic, considering that the cause
is an act of language initiated by one character in the novel for another.
239
The pragmatic discursive function is also well represented for the

motif read_DET_letter of the corpus GEN EN: the motif occurs within
a dialogue between two characters and is part of a language act directly
oriented towards the addressee (however, not all occurrences of read_
DET_letter in the direct speech have a pragmatic discursive function).
(14) Abruptly Fritz leaned towards her and said, “You know that I have
only one thing to ask. Has he read my letter? ” (Fitzgerald The Blue Flower,
1988)
The motif here is the heart of an interrogative sentence, whose illocu-

tionary status is indeed the request (from a son to his mother). The
answer to the question is important for the character, and it will have an
impact on the rest of his actions.
Thus, for lire_DET_lettre and read_DET_letter, the discursive func-
tion is often narrative because reading the letter allows the action to
progress, either directly or through interactions that will then advance
the action; for GEN FR, reading the letter sometimes allows a character
to be freeze-framed for a description. One does not find here the meta-
fictional temptation of the literary novel. However, people read more
letters in GEN than in CRIM (and, presumably, than in other subgen-
res): these novels are more dominated by the written word, and by the
intimate interactions between the characters, some of which go through
the letter exchange.
Lire_DET_livre and read_DET_book

There are more occurrences of this motif in GEN (FR and EN both)
than of the previous ones. Clearly, since the 1950s, we have been read-
ing fewer and fewer letters, and receiving more and more calls and mes-
sages. An excerpt from the GEN EN corpus shows that: “It’s a weird
thing to do, isn’t it, writing a letter in these days of t’internet!” (Nicholls
One Day, 2009). Moreover, as mentioned above, since the reflective
writing of the novels in the GEN corpus is quite specific, there will be
more mises en abyme of the reading of a book. The book is not addressed
like the letter: the latter establishes links between the characters, it is
240
J. Sorba et al.
written by one of them for another, and as such very often has a nar-
rative function. Reading a book, on the other hand, is a solitary activ-
ity (such as writing) and willingly free (in terms of narrative), which
encourages introspection rather than action.
It is not the title of the book that matters: for GEN FR, more than
half determiners of book are indefinite (57.5%). And in French, the
reference of the read letters is more often determined (by the definite
article and the complements) than for this motif. Thus reading a book
would be more of a pose than a narrative element advancing action. In
fact, the simple past is less used than to lire_DET_lettre and the discur-
sive function of the motif is less often narrative: reading a book mainly
disrupts the reader’s inner life, while the letter also has an impact on his
actions.
(15) … lisant, à quelque temps de là, un livre dont l’auteur, à un détour

de page, disait qu’il regardait, “à l’horizon, comme un point qui grossissait”,
je m’étais promis que si j’écrivais un jour, plus tard, jamais je n’userais,
dans les livres que j’affabulerais à dessein d’un grand texte efflorescent, de
ces expressions qui avaient été, à leur naissance (et dans une rhétorique
naïve), aiguës, mais que la répétition, au fil des siècles, avait émoussées.
(Berger Les Matins du Nouveau Monde, 1987) (…reading, at a certain time
from there, a book whose author, at a detour, said that he was looking “at the
horizon, like a point that was growing”, I promised myself that if I wrote
one day, later, I would never use, in the books that I would intentionally
elaborate with a large efflorescent text, those expressions that had been, at
birth (and in naïve rhetoric), sharp, but that the repetition, over the cen-
turies, had blunted.)
Here the discursive function is narrative (in terms of inner life), since
this reading advances the narrative of the literary vocation: the narrator
chooses to take a different path from the author of the book he men-
tioned for the work of his style. The EN corpus also presents occur-
rences of this motif with a narrative discursive function, quite similar to
the FR corpus, but the narrative discursive function remains much less
used than lire une lettre and read a letter.
241
For this motif, a cognitive discursive function could be found, which

is not the most represented either:
(16) Il n’aime pas la musique. Il trouve que ça gêne. Que c’est une servi-
tude. Parce que, quand on en écoute, on est obligé d’entendre le morceau
sans l’interrompre, alors que quand on lit un livre, on peut le lire dans le
désordre, sauter des pages, que la liberté est totale. Il aime cette liberté et
ne supporte pas d’en être privé. (Angot Une semaine de vacances, 2012)
(He doesn’t like music. He thinks it disturbs him. That it’s a servitude.
Because, when you listen to it, you have to hear it without interrupting
the song, whereas when you read a book, you can read it in disorder, skip
pages, that freedom is total. He loves this freedom and cannot stand to be
deprived of it.)
The example starts off as descriptive: it describes the tastes of the male
character. However, the motif in its minimal form is introduced by an
indefinite subject (on ): it belongs more explicitly to the reported speech
of this same character, who sets out his opinion. The subordinate clause
quand on lit un livre is a circumstance to which he attaches a practice:
the motif has a cognitive discursive function, it is a reflection formu-
lated by the character. The same motifs can be found in the EN corpus:
(17) Will I ever be able to read a book again? Will I ever be able to talk
like old friends with my wife? (Parks Europa, 1997)
In this case, the discursive function of read a book is cognitive: it inter-

venes in a projection, a hypothesis of the narrator. It is only fantastically
part of the fiction action, but is also a part of the inner discourse. In the
EN corpus, the DF of this motif is as often pragmatic as it is cognitive
and narrative:
(18) “But I’m bored. Can’t I come?”

“Not this time,” says my mother. “Read a book, or watch the television.”
(Hyland Carry Me Down, 2006)
242
J. Sorba et al.
The motif takes an imperative form: it reflects the speech act—order or

advice given by the mother to her son (the narrator). But this pragmatic
DF is less present, relatively, in the FR corpus, even if it is encountered
there; it is doubtful that this disparity is really significant, because in
both EN and FR corpora, the DF of the lire_DET_livre and read_
DET_book motifs is more massively descriptive. In direct speech, char-
acters describe themselves as readers, or are described as never reading
a book; in the story, their habits are described (lying down with a book
for example) especially if the character’s profession is to wait (as a guard,
a driver). In the following example, the narrator climbed on a bus:
(19) Je m’étais assis en face d’une fille qui lisait un livre. (Laurent Les Sous-
Ensembles flous, 1981) (I had sat across from a girl reading a book.)
The motif occurs in a relative clause epithet of the name, which char-
acterizes it. The imperfect tense lisait is secant, the action has no final
limit: thus the reading lends itself to description (the narrator tries to
know what she is reading, and the girl is physically described on this
occasion). Very similar examples can be found in the EN corpus:
(20) The interior was expensively, even tastefully furnished. At a round

table in the centre of the room, reading a book, sat Marjorie. (Storey
Pasmore, 1972)
(21) Wherever it was I met Ceel, she was always there, always waiting,
almost always reading a book – usually something recent I’d heard of.
(Banks Dead Air, 2002)
A character busy reading (the verb has a durational aspect in its form
lisait and reading ), whether a letter or more willingly a book (which
encourages more immobility by its length), is a subject often taken up
in painting for example; the novel of literary fiction, which shows a
greater reflection on the practice of writing, also uses more cultural ref-
erences than can be found in the CRIM corpus. The representation of
reading shows these two tendencies: the importance given to the written
word and the claim to legitimate culture.
243
4.3.2 Writing as Motif
Écrire_DET_lettre and write_DET_letter

The results presented in Gonon et al. (2018) for the motif écrire une let-
tre in GEN FR can be compared to the results for the motif write a let-
ter in the GEN EN corpus.
This motif in GEN EN has a discursive function that varies accord-
ing to the context: it can be descriptive, when the motif makes it pos-
sible to evoke the behaviour of a character (who does not write letters,
who writes them in such and such a way), but this descriptive discur-
sive function concerns relatively few occurrences; it occurs a little more
often in the GEN FR corpus, without this difference being really signif-
icant. Pragmatic discursive function is fairly well represented in English
(and a little less so in French), particularly when it appears in direct
speech. Pragmatic discursive function then readily fits in with the use
of the personal pronoun tu as the subject of the verb and with inter-
rogative sentences. The write_DET_letter motif is part of a speech act
directed at the addressee, another character in the novel:
(22) “Will you write a letter for me?

“What sort of a letter?”
“A crook letter. From Charles Morgan to myself.” (Spark The Girls of
Slender Means, 1963)
The speaker, an author, proposes to the young collaborator of the pub-

lisher Charles Morgan to write to the latter a very enthusiastic bogus
letter from a fictional publisher, so that Morgan will offer a more advan-
tageous contract to the speaker. The discursive function is pragmatic: it
makes it possible to carry out an act of language (the request for a ser-
vice) that links two characters in the story. It is less used in the GEN FR
corpus, probably because the French GEN novels allow a more descrip-
tive and reflective realization of this motif (that is, writing a letter can
portray a writer’s character as having a particular affinity with the act of
writing).
244
J. Sorba et al.
Like écrire_DET_lettre in GEN FR, the motif in the GEN EN cor-

pus is very often captured on an aspectual or modal level: its discur-
sive function is often cognitive in both corpora. It reflects assumptions
or projections by a character, who imagines that he is writing a let-
ter, or wondering who wrote it, et cetera. In the example below, Mr.
Bhoolabhoy, manager of the hotel owned by his wife Lila, must give the
brush-off to the Tusker couple, but he appreciates the husband:
(23) Writing the letter would put the seal on his total and abject surrender
to Lila. He knew he had already surrendered. But it was still a difficult
letter to write. It was like composing a warrant for the execution of an old
friend. (Scott Staying On, 1977)
The motif is linked to what the act of writing the letter means for the
character, the consequences he projects for the rest of the action.
However, the most represented of the discursive functions is narra-
tive, as is the case for the equivalent motif of the FR corpus:
(24) He himself would need from Schlinker details of where the arms
shipment was supposed to be heading, so that the captain could draw up
the appropriate manifest. … He wrote a long letter to Mr. Stein as chair-
man of Tyrone Holdings, instructing him to prepare the papers for a
board meeting of the company in his office four days hence, with two
resolutions on the agenda. (Forsyth The Dogs of War, 1974)
Cat Shannon, a mercenary, organizes in this passage the delivery

of weapons and ammunition, via a holding company he created in
Luxemburg. The motif is part of a series of actions performed by the
character in order to recover the cargo of firearms.
Thus, when it comes to écrire une lettre and write a letter, pragmatic
and narrative discursive functions reflect the fact that the letter plays
a role in the action (such as reading it) or is the object of interactions
between characters. Cognitive function itself may correspond to the
course of the plot, since a character wonders what it means to write a
letter to another character: when the writing is transitive (directed at
someone or something), relatively fewer metafictional considerations are
involved.
245
Écrire_DET_roman and write_DET_word

Here we can distinguish the two GEN corpora FR and EN: as quantita-
tively equivalent as the previous motifs turned out to be in the two fiction
traditions French and English, écrire_DET_roman seems much more char-
acteristic of GEN FR. Could a possible hypothesis be that it is due to the
French literary novel being more metafictional than the EN literary novel?
To briefly recap the results presented in Gonon et al. (2018) for the
GEN FR corpus: the motif écrire_DET_roman occupies a descriptive
discursive function (the character presents himself as a writer, for exam-
ple), at other times a narrative discursive function (the action of writing
a novel is inserted in the plot). But, the altogether characteristic discur-
sive function of this corpus is a so-called commentary function, one
indicating a reflexive approach to writing that takes place in a pause in
the narrative (a writer then talks about the writing process). This motif
hardly appears in the GEN EN corpus.
Nevertheless, write_DET_word is representative of it, and its discur-
sive function is more narrative. Its context is indeed much less metafic-
tional than for the motif écrire_DET_roman:
(25) When nobody answered, she turned to Jetta: “Perhaps Roddy knows
the answer.” Jetta looked at me and then replied, “I’m sorry. Roddy does
not know and nor do I.” Miss Galbraith looked disappointed and turned
to write the word on the board. (Burnet His Bloody Project, 2016)
Since student Roddy does not know the answer—according to Jetta—the

teacher acts accordingly. The motif enters a narrative chain, in the broader
sequence of the geography course. Here the two corpora frankly diverge.
4.3.3 Conclusions About the Discursive Functions of Reading

and Writing as Motifs Comparing GEN FR and GEN EN
Points in Common
For the motifs écrire/lire_DET_lettre and read/write_DET_letter, the
most representative discursive function is often narrative. The narrative
function is also present for the other realizations of the motifs built with
lire/read and écrire/write, which is no surprise for the novel. Write and
246
J. Sorba et al.
read denote actions that advance the plot: in this case, writing and read-
ing are transitive, they cause an action and integrate themselves in a nar-
rative causality chain.
For these reasons, pragmatic and cognitive discursive functions reflect
a specificity of GEN, which very often deal with the relationships
between the characters, their intimate life and what they say about it.
Thus letters and books become objects of discussion which intensify
the questions and assumptions of the characters and the actual written
interactions that the letter implies.
Finally, the descriptive discursive functions for lire_DET_livre and
read_DET_book show the mise en abyme of the reading act, quite typical
of reflective GEN writing (the reader reads a book that also features a
reader). It further reflects GEN’s tendency to represent a character in
the process of reading and to institute reading as a descriptive attrib-
ute: to say that one reads or does not read is to give an image of one-
self and others in which reading becomes a value—to which one either
subscribes (it is positively connoted) or which one derogates (the judge-
ment is then pejorative).
Characteristics of Each GEN Corpus

In general, the PhraseoRom’s GEN FR corpus is much more reflec-
tive than the GEN EN corpus: it features more writers, more mises en
abyme of reading and writing practices. Its metafictional dimension is
more important. Foremost, this shows the importance of the pragmatic
discursive function in GEN EN for read_DET_book and write_DET_
letter: these activities are perceived as more static (descriptive and cog-
nitive discursive functions) in GEN FR and more dynamic in GEN EN
(as the object of verbal interactions). Then, lire_DET_lettre performs
a more descriptive discursive function in GEN FR; in GEN EN, the
discursive function of read_DET_letter is more likely to be narrative or
pragmatic: the same conclusions can be drawn.
Lastly, the motifs constructed around écrire/write diverge statistically
in both corpora: write_DET_word is specific to the GEN EN corpus,
and it has a mainly narrative discursive function: no reflective dimen-
sion here. But écrire_DET_roman is significant for GEN FR: the motif
247
sometimes has a commentary discursive function specific to GEN (and

more precisely to GEN FR) since it introduces a reflexive pause into the
plot, letting the character wonder or discuss what writing is or projects a
novel. The mise en abyme is more typical of the GEN corpus, and espe-
cially of GEN FR, for what concerns not only the reader but above all
the novelist himself: this could explain the preference for the verbal first
person for realizing the motif écrire_DET_roman, or even for the other
motifs built around lire and écrire, which also have a connection with
the written language.
5 Conclusion
The data used in this study are based on the contrast of phraseologi-
cal elements in two genres, the crime novel and literary fiction. Crime
novels were solely chosen for statistical reasons to serve as contrasts to
the novels of literary fiction. We then studied the statistically signifi-
cant LSC that included the verbs écrire and lire (une lettre, un roman/un
livre ) for French and read and write (a letter, a word/a book ) for English.
Our study therefore contrasted the English and French PhraseoRom
corpora of the two literary genres crime novels and literary fiction to
answer two research questions: Would the constructions around lire/
read and écrire/write allow us to define the novel of literary fiction? And
how are motifs of the French and English corpus which centre around
these verbs characteristically realized? We were able to establish that écri-
re/lire_DET_lettre, write/read_DET_letter and écrire/lire_DET_roman,
write_DET_book and write_DET_word are indeed motifs. Since these
significant LSCs show many syntagmatic and paradigmatic variations,
we can designate them as motifs for more detailed analysis. The var-
iations of the discursive functions and the role the motifs play in the
literature show the richness of their stylistic realizations. The statisti-
cal significance of these LSCs (compared to crime novels) shows that
reading and writing is specific in literary fiction (GEN). The discursive
functions of motifs are narrative and descriptive but other functions
(cognitive, pragmatic, and commentary) highlight the importance of
248
J. Sorba et al.
this subject to an author—and for the characters portrayed as readers

and writers. The reading and writing of a letter, of a book or of words
may be transitive (directed from one person to another) but one of
the characteristics of the French GEN corpus is “intransitive” writing
(Barthes 2002), thus reflecting an author’s own practice.
References
Adam, Jean-Michel. 2011 (1992). Les Textes: types et prototypes, 3rd ed. Paris:
Armand Colin.
Austin, John Langshaw. 1970. Quand dire, c’est faire. Translated by Gilles Lane.
Paris: Seuil.
Barthes, Roland. 2002 (1970). “Écrire, verbe intransitif?” In Œuvres Complètes,
1968–1971, Vol. 3, edited by Éric Marty, 617–26. Paris: Seuil.
Biber, Douglas. 1993. “Using Register-Diversified Corpora for General
Language Studies.” Computational Linguistics 19: 219–41.
Blumenthal, Peter. 2006. “De la logique des mots à l’analyse de la synonymie.”
Langages 150 (2): 14–31.
Dunning, Ted. 1993. “Accurate Methods for the Statistics of Surprise and
Coincidence.” Computational Linguistics 19 (1): 61–74.
Sorba. 2018. “Motifs textuels spécifiques au genre policier et à la littérature
‘blanche’.” In 6e Congrès Mondial de Linguistique Française, edited by Frank
Neveu, Bernard Harmegnies, Linda Hriba, and Sophie Prévost. SHS Web
of Science. https://www.shs-conferences.org/articles/shsconf/pdf/2018/07/
shsconf_cmlf2018_06007.pdf.
Gonon, Laetitia, and Julie Sorba. Forthcoming. “Phraséologismes spécifiques
dans les romans historiques et les romans de littérature blanche.” Journal of
French Language Studies.
Grossmann, Francis, Salah Mejri, and Inès Sfar. 2017. “Présentation.
Phraséologie: sémantique, syntaxe, discours.” In La Phraséologie: sémantique,
syntaxe, discours, edited by Francis Grossmann, Salah Mejri, and Inès Sfar,
7–12. Paris: Honoré Champion.
249
Kraif, Olivier. 2016. “Le Lexicoscope: un outil d’extraction de séquences

phraséologiques basé sur des corpus arborés.” Cahiers de lexicologie 108 (1):
91–106.
Legallois, Dominique, and Agnès Tutin. 2013. “Présentation: Vers une exten-
sion du domaine de la phraséologie.” Langages 189 (1): 3–25.
langue au discours.” Langages 189 (1): 65–79.
Mahlberg, Michaela. 2007. “Clusters, Key Clusters and Local Textual
Functions in Dickens.” Corpora 2 (1): 1–31.
Malrieu, Denise, and François Rastier. 2001. “Genres et variations morphosyn-
taxiques.” Traitement automatique des langues 42 (2): 547–77.
Martin, Robert. 1983. Pour une logique du sens. Paris: PUF.
Novakova, Iva, and Julie Sorba. 2018. “La Construction du sens autour des
lexies d’affect: proposition d’un modèle fonctionnel.” Langages 210: 55–70.
Rastier, François. 2011. La Mesure et le grain. Sémantique de corpus. Paris:
Honoré Champion.
Patterns in Post-war Fiction.” Functions of Language 22: 362–99.
Siepmann, Dirk. 2016. “Lexicologie et phraséologie du roman contemporain:
quelques pistes pour le français et l’anglais.” Cahiers de lexicologie 108 (1):
21–41.
Sinclair, John. 1991. Corpus Concordance Collocation. Oxford: Oxford
University Press.
Sinclair, John. 2004. Trust the Text: Language, Corpus and Discourse. London:
Routledge.
Tutin, Agnès, and Olivier Kraif. 2016. “Routines sémantico-discursives dans
l’écrit scientifique de sciences humaines: l’apport des arbres lexico-syntax-
iques récurrents.” Lidil 53: 119–41.
Zaharieva, Radostina, and Stefana Kaldieva-Zaharieva. 2017. “Des principaux
termes dans le domaine de la phraséologie.” In La Phraséologie: sémantique,
syntaxe, discours, edited by Francis Grossmann, Salah Mejri and Inès Sfar,
15–37. Paris: Honoré Champion.
9
Dans un état de NP and in a state
of NP: Bridging the Syntagmatic Gap
in English and French Fiction
Susanne Dyka, Ludwig Fesenmeier
and Marion Gymnich
1 Introduction
In checking the lexicographic description of the meaning of nouns
designating states against their collocational behaviour in authentic

French newspaper texts (Le Monde 1994), Peter Blumenthal points out
that certain nouns do not simply combine with prepositions (such as
“Dans la solitude, on a davantage de respect pour les autres” [In solitude,
one has more respect for others1] or “être dans l’embarras” [be in trouble]).
S. Dyka (*) · L. Fesenmeier

L. Fesenmeier
e-mail: ludwig.fesenmeier@fau.de
M. Gymnich
https://doi.org/10.1007/978-3-030-23744-8_9
252
S. Dyka et al.
Instead, they occur in combination with lexical items such as état (state),
moment (moment) or période (period), for example: “mort dans un état de
déchéance physique extrême” (died in a state of extreme physical degenera-
tion); “Primo Levi écrivait dans un moment de découragement” (Primo Levi
wrote in a moment of discouragement); “dans une période de crise, les com-
pagnies ont accueilli favorablement l’A340” (in a period of crisis, the com-
panies have welcomed the A340) (see Blumenthal 2004, 148, 152–53).
This phenomenon assumes particular interest for us when the noun
combined with état, moment or période itself designates, at least accord-
ing to the lexicographic definition, a state or a certain lapse of time: in
the Trésor de la Langue Française informatisé, for example, crise (crisis) is
defined as “[m]anifestation brusque et intense, de durée limitée ” (sense
I.B; emphasis added) (sudden and intense manifestation, of limited
duration ) and confusion (confusion) as “[é ]tat de ce/celui qui est confus”
(senses A/B; emphasis added) (state of something/somebody which/who
is confused). One should thus expect—and indeed finds—examples such
as in (1) below but also where the “state” can be made explicit, as in (2).
(1) L’idée que mon siège de DS s’était vu transpercé était elle aussi très
douloureuse, et achevait de me plonger dans une confusion mentale rageuse.
(GEN, Chamoiseau Hyperion victimaire: Martiniquais épouvantable,
2013) (The thought that my DS seat had holes in it equally hurt very
much and managed to throw me into a furious mental confusion.)
(2) Cela ne dura guère, mais l’expérience me laissa dans un état de con-
fusion mentale absolue. (FY, Héliot Reconquérants, 2001) (It did not last
long, but the experience left me in a state of total mental confusion.)
While temporal expressions such as moment or période seem to be used

especially for expressing the limited duration of the state or offering a tem-
poral reference point for a given event (see Blumenthal 2004, 153), the
use of dans un état de NP is explained by the widespread tendency to add
this prepositional phrase as a way of bridging a kind of syntagmatic gap.2
2Obviously, the expression “syntagmatic gap” is not to be confused with the term coined by Teich
(1999) and defined as the “lack of means to express generalizations about syntagmatic structures
and their morphosyntactic properties” (1999, 221).
9 Dans un état de NP and in a state of NP: Bridging …
253
This is the kind of gap that seemingly opens when the aim is to designate
the final state or at least an advanced stage in an ongoing process, or, more
generally, when trying to evoke the idea of a paradigm of different steps
considered to be relevant in such a process (see Blumenthal 2004, 152).3
Nevertheless, while such an explanation can account for examples such as
(3), it disregards constellations such as (4), which invoke a rather meta-
phorical reading of sommeil (sleep).4
(3) Il plongea dans un sommeil profond et retrouva cet atelier imaginaire

qu’il avait créé selon son bon vouloir. (FY, Gaborit Les Chroniques des
Féals 2: Le Fiel, 2001) (He fell into a deep sleep and returned to this imagi-
nary studio that he had created completely on his own.)
(4) La vision romane était différente, bien plus unanimiste: à sa mort le
croyant entrait dans un état de sommeil profond, et se mêlait à la terre.
(HIST, Houellebecq Soumission, 2015) (The Romanesque vision was
much more communal: at his death, the believer fell into a deep sleep and
was laid in the earth [Houellebecq 2016, 136].)
In this article, however, we will not concern ourselves with possible dif-
ferences between dans NP and dans un état de NP, but instead do a more
detailed analysis of the occurrences of dans un état de NP as well as its
English counterpart in a state of NP in French and English literary texts
across different subgenres. More precisely, while Blumenthal’s results
from his analysis of a newspaper corpus appear to be confirmed by an
analysis of French literary texts (see example 2), examples such as (4)
suggest that dans un état de NP may also fulfil other functions.
3Blumenthal (2004, 152) points out that this is true of the majority of the occurrences of dans
un état de NP; indeed, there are only very few examples which do not fit the case, see for exam-
ple “La centrale d’achat d’espace Carat … est la plus lourdement frappée, … pour avoir notam-
ment placé les supports journaux, radios, télévisions dans un état de dépendance économique ” (The
advertising agency Carat is the one most affected, since it has put in a state of economic dependency
in particular the newspapers, radio and TV stations). In this case, the adjective économique speci-
fies the domain for which the état de dépendance is asserted.
4According to the Trésor de la Langue Française informatisé, sommeil is an “[é]tat dans lequel se
trouve un être vivant qui dort” (sense A.1; emphasis added) (a state in which a living being that
sleeps finds itself ) or an “[é]tat d’inertie, d’inactivité momentané où se trouve quelque chose”
(sense B.1; emphasis added) (a state of indolence, of momentary inactivity in which something
finds itself ).
254
S. Dyka et al.
Therefore, since the expressions dans un état de NP/in a state of NP

are not only “an ordered subset of the textual ensemble, formed by
the recurring combination of n elements” (Longrée and Mellet 2018,
156), but also fulfil certain identifiable functions on the semantic, prag-
matic and rhetorical levels (see Longrée and Mellet 2013, 66), we treat
them as motifs (see Chapter 2 by Legallois and Koch in this volume).
Their concrete manifestations and functions in the different French and
English literary sub-corpora will now be investigated.
2 The Motifs dans un état de NP and in

a state of NP in the Corpus: Statistical
Findings
We begin with a look at the frequency and collocates of dans un état
de NP and in a state of NP in the different literary sub-corpora of the
PhraseoRom project. Using Lexicoscope’s “lexicogram” tool (see below),
we located statistically significant collocates of dans un état de NP and
in a state of NP in the respective languages and corpora (Sect. 2.1). The
findings will then be analyzed in more detail (Sect. 2.2).
2.1 Frequency of dans un état de NP and in a state

of NP and the Number of Statistically Relevant
Collocates
As a first step, we analyzed the frequency of the pattern dans un état de

NP in the French sub-corpora and of in a state of NP in the English
sub-corpora (tokens) as well as the number of statistically relevant col-
locates (types). The pattern for dans un état de NP corresponds to the
search represented by the Recurrent Lexico-Syntactic Tree (RLT) shown
in Fig. 1; the pattern for in a state of NP corresponds to a search for the
RLT shown in Fig. 2.
As some of the sub-corpora differ quite considerably in size, the fre-
quencies were normalized to make the results comparable (see Appendix
A for the corpora). A collocate is considered to be statistically relevant if
255
Fig. 1 Recurrent Lexico-Syntactic Tree for dans un état de NP in the French

sub-corpora
Fig. 2 Recurrent Lexico-Syntactic Tree for in a state of NP in the English

sub-corpora
its log-likelihood ratio (LLR) is greater than or equal to 10.83. In addi-

tion, we assigned a value of “3” as the threshold of dispersion, meaning
that the statistically significant co-occurrences must be found in at least
three different texts. The results are shown in Table 1.
The statistically relevant collocates were identified by the above-
mentioned co-occurrence analysis with Lexicoscope. It produces “lex-
icograms” that list the most frequent collocates of a given node word
or expression “weighted by frequency values (for occurrences and
256
S. Dyka et al.
Table 1 Frequency of dans un état de NP and in a state of NP with statistically

relevant collocates and number of statistically relevant collocates (LLR ≥10.83)
Sub-corpus French English
Frequency Number of sta- Frequency Number of statisti-
per million tistically relevant per million cally relevant collo-
words collocates (tokens) words cates (tokens)a
CONT 0.79 47 3.75 267
CRIM 3.30 59 10.39 76
FY 3.45 46 9.32 111
GEN 5.83 200 15.26 210
HIST 12.16 138
ROM 2.75 27 5.60 48
SF 6.91 91 5.86 49
aThe lexicogram included punctuation marks in the collocates, but these were
not included in the number of collocates. Punctuation marks and what they
might signify will be discussed below (see Sect. 2.2.4)
co-occurrences) and association measures” (Kraif and Diwersy 2014,

384; see also Kraif and Diwersy 2012, 400). We generated lexicograms
for the motifs dans un état de NP and in a state of NP for every subgenre
to give us an overview of the most relevant collocates occurring with the
motifs. This proved helpful as it revealed the occurrences of a number
of different nouns, adjectives and verbs with the motif dans un état de
NP/in a state of NP. The results of this co-occurrence analysis will be
discussed in more detail in the next section, but Table 1 already shows
interesting differences between the subgenres: for French, the subgenres
crime, fantasy, general and science fiction show the highest number of
statistically relevant collocates, while these are completely absent from
historical fiction. By contrast, in English the subgenres general, fantasy
and historical fiction provide the highest number of statistically relevant
collocates. The genre with a high count of relevant collocates in both
the English and French corpus therefore turns out to be general fiction.
When we compare English and French, as in Table 1, we find that in
all subgenres except for science fiction the construction in a state of NP
occurs with greater frequency (normalized) in the English sub-corpora
than the French expression dans un état de NP does in the French
sub-corpora. At first glance, this might be at least partly due to the fact
that in French there also exists en état de NP (literally “in state of ”;
257
Table 2 Frequency of en état de NP

Sub-corpus Number of occurrences (tokens) Frequency per million words
CONT 30 0.50
CRIM 20 1.12
FY 5 0.38
GEN 20 0.58
HIST 5 0.34
ROM 2 0.20
SF 14 1.06
see Table 2) alongside dans un état de NP for expressing more or less spe-
cific psychological states such as depression, shock or dementia (être en
état de depression/choc/démence [be in a state of depression/shock/demen-
tia]; see Blumenthal (2004, 153)). The English equivalent would also be
in a state of NP (for example, in a state of shock), since in English there is
no alternative construction like en état de NP.5 Nevertheless, in the final
analysis, en état de NP turns up more frequently in expressions such as en
bon/mauvais/parfait état de fonctionnement/marche (in good/poor/perfect
working order), which would not be rendered by *in a good/poor/perfect
state6 in English. These differences in frequency between the French and
the English sub-corpora need further investigation.
2.2 Statistical Analysis of dans un état de

NP and in a state of NP
The statistically significant collocates (types) that occur with the motifs
in a state of NP and dans un état de NP are shown, respectively, in
Tables 3 and 4. The findings here are grouped by word classes, allowing
us to compare relevant collocates in the different subgenres.
5An alternative in English would be be in shock, which does not contain any “linking” element.
6Very rarely, one finds also examples such as “Exceptionellement, parce qu’il se sentait en léger état
de faiblesse, Marc accepta de déroger au travail” (CRIM, Vargas Un peu plus loin sur la droite, 1996)
(Exceptionally, because he was feeling rather weak, Marc agreed to put aside his work [Vargas 2014,
102]); see, however, “… je me sentais dans un état de faiblesse épouvantable …” (GEN, Djian Zone
érogène, 1984) (… I found myself in a state of terrible weakness …).
258
S. Dyka et al.
Table 3 Statistically significant collocates of in a state of NP sorted by word

classes
Sub-corpus CONT CRIM FY GEN HIST ROM SF
Collocates’ word class
Verb 5 2 2 6 2 1 4
Noun 1 1
Adjective 2 5 5 3 2
Quantifier 1 1
Conjunction 1 1
Punctuation mark 2 2 2 2 3 1 2
Total 8+2 3+2 8+2 16 + 2 6+3 3+1 4+2
Table 4 Statistically significant collocates of dans un état de NP sorted by word

classes
Collocates’ word class
Verb 1 1 6 2
Noun 4 6 5 20 2 5
Adjective 2 1 2 2 4
Total 5 8 7 28 4 11
As can be seen in Tables 3 and 4, the range of collocates is widest in gen-

eral fiction for both languages. In the English sub-corpora, verb collocates
appear in all sub-genres. Only one noun collocate was found in English
general fiction.7 The highest number of adjective collocates can be found
for GEN and FY; for HIST and ROM, respectively three and two adjec-
tive collocates exist. In SF and CRIM, no adjective collocates were specific.
In the French sub-corpora, noun collocates are prevalent in all
sub-corpora except ROM, while the most verb collocates appear in
GEN but are completely absent from CRIM and ROM. Adjectives
turn out to be rare, except in SF, which contains the highest number of
adjective collocates.
Some findings are specific to the English corpora: one quantifier
(some ) was found for GEN and FY and one conjunction (and ) seems to
7The second noun collocate that was found was near. It seems to be wrongly classified by the parser:
in all cases it is an adjective (near collapse, near exhaustion, near mutiny) premodifying the noun that
occurs with in a state of. Therefore near was not counted as noun collocate.
259
occur quite often in CRIM. Punctuation marks were identified as rele-

vant “collocates” for all genres—although their numbers differed. These
findings will be discussed in Sect. 2.2.4.
2.2.1 Verb Collocates of dans un état de

Turning to a more detailed analysis (see Table 5) of the statistically

significant verbal collocates of dans un état de NP, we learn that verbs
denoting a change of state clearly predominate, in particular plonger
dans un état de NP (see examples (5a/b)), while the simpler “being in a
state” (trouver/vivre dans un état de NP, see examples (6a/b)) is much less
in evidence. However, the subgenres differ when it comes to verbs used
to express a “change of state” versus “retention”: trouver/vivre dans un
état de NP, which denote “being in a state”, are statistically significant
only in GEN, while “change of state” is expressed by mettre, plonger and
tomber; the only verb present in both FY and SF is plonger, which is also
statistically most significant overall. The verb laisser is found only in SF
and with weak statistical relevance.
(5a) Cette léthargie mystérieuse avait plongé la cour des Burgondes dans
un état de profonde perplexité. (FY, Brasey La malédiction de l’anneau,
2008) (This mysterious lethargy had plunged the Burgundy court in a
state of deep perplexity.)
Table 5 Statistically significant verb collocates of dans un état de NP (LLR

≥10.83)
Collocate
être (be) 12.64
laisser (let) 13.72
mettre (put) 19.01
plonger (plunge) 43.87 81.34 105.46
retrouver (find) 19.64
tomber (fall) 15.43
trouver (find) 21.50
vivre (live) 20.27
260
S. Dyka et al.
(5b) … je plonge dans un tel état de mélancolie que je réalise que je n’ai dû
faire que ça … (GEN, Jenni L’Art français de la guerre, 2011) (I fall into
such a state of melancholy that I realize that I had to do just that.)
(6a) Mais ce soir-là, il se trouvait dans un état de lassitude extrême. (GEN,
Sabatier Le roman d’Olivier 1: Les Allumettes suédoises, 1969) (But that
evening, he found himself in a state of extreme tiredness.)
(6b) Je vécus durant de nombreuses semaines dans un état de félicité et de
douceur et d’excitation et d’évidence et de luminosité que je n’avais jamais con-
nues si longtemps. (GEN, Quignard Le Salon du Wurtemberg, 1986) (I lived
for many weeks in a state of bliss and of placidity and of agitation and of evi-
dence and of radiance which I had never known to last for such a long time.)
In the English subgenres, unlike in the French corpora, be is significant

as a collocate for every genre (see Table 6). ROM is the only genre in
which be alone seems to be a significant collocate of in a state of. This
may simply result from the fact that be is the most common collocate
of in a state of in general; using the verb be is presumably the most gen-
eral or “neutral” way of embedding in a state of into a sentence. In this
construction, the focus of interest is the NP following in a state of and
not the verb. In CRIM, FY and HIST two verb collocates were found:
be in all three and either keep or live (see example (7)). SF shows four
verb collocates, three of which are not specific to the other genres: leave,
remain, and seem. GEN seems to have the widest variety of specific verb
collocates, some of which do not seem to be specific in the other gen-
res, for example exist, find and spend. In contrast to our findings for the
Table 6 Statistically significant verb collocates of in a state of NP (LLR ≥10.83)

Collocate
be 114.16 73.60 31.26 75.05 85.79 34.73 19.77
exist 26.40
find 17.01
keep 15.55 11.05 14.76
leave 44.07 15.47
live 114.42 35.51 46.41 41.17
remain 39.98 35.83
seem 13.96
spend 27.83
261
French corpus, all of the English verbs seem to express remaining in a

specific state, even leave, which always occurs in the construction exem-
plified by (8).
(7) It will keep yesterday’s memories in a state of flux for a few hours. (FY,
Pratchett The Dark Side of the Sun, 1976)
(8) They’d also left the room in a state of complete disarray; the whole place
had clearly been picked over for evidence. (FY, Barker Galilee, 1998)
2.2.2 Noun Collocates of dans un état de

In looking at the statistically significant nouns occurring with dans un

état de NP (see Table 7), the presence of conscience, nerf and légume may
at first seem surprising from an onomasiological point of view. But a pat-
tern emerges when we look at where they occur: conscience appears most
often in expressions like dans un état de semi-conscience (in a state of inter-
mittent consciousness) and nerf is always used in the plural, for example,
in dans un état de nerfs épouvantable (in a dreadful state of nerves). Finally,
as the entry “légume” in the Petit Robert suggests, état de légume (state
of vegetable) indirectly connotes a state, since as a calque on the English
vegetable,8 légume can also designate a “[m]alade dans un état végétatif
chronique” (ill person in a chronic vegetative state) (légume, sense III):
(9) … mon père est resté dans un état de légume pendant 26 ans dans un
hôpital. (CONT, Toute une histoire ) (… my father remained in a vegetable
state for 26 years in a hospital.)9
The crime novels show a certain clustering of nouns related to nervous-

ness (tension, nervosité, nerfs ), while science fiction novels seem to be
characterized instead by descriptions of physical exhaustion (faiblesse,
8See OED, “vegetable”, noun, sense 2 (b): “A person likened to a plant, spec. … one who is incapa-
ble of normal mental or physical activity, esp. as a result of brain damage”.
9See below, Sect. 3, for further discussion.
Table 7 Statistically significant noun collocates of dans un état de NP (LLR ≥10.83)
Sub-Corpus CONT CRIM FY GEN HIST ROM SF
262
Collocate
coma (coma) 67.14
conscience (consciousness) 112.69 76.00 102.15
crise (crisis) 41.50
S. Dyka et al.
délabrement (decay) 132.65 300.12 269.67

détresse (distress) 70.08 47.90
faiblesse (weakness) 100.24 86.63
fatigue (tiredness) 87.11 64.19 55.12
fébrilité (feverishness) 94.30
fraîcheur (freshness) 66.00
légume (vegetable) 66.98
manque (lack) 72.26
nerf (nerve) 109.28 97.82
nervosité (nervousness) 73.18 71.59
panique (panic) 63.20 59.16
perplexité (perplexity) 90.62
prostration (prostration) 88.77 125.58
rage (rage) 138.23
saleté (dirt) 92.83
stress (stress) 71.09
stupeur (stupor) 85.44 97.04
tension (tension) 56.06 60.25 105.01
tristesse (sadness) 47.58
263
Table 8 Statistically significant noun collocates of in a state of NP (LLR ≥10.83)a

Collocate
shock 64.25
aSee Footnote 7 with regards to near
fatigue, prostration ). The preference in fantasy fiction seems to be for

nouns indicating a kind of helplessness coupled with a negative form of
excitement, such as fébrilité, panique and perplexité:
(10) Le sire Marten était de plus en plus nerveux. … Il était dans un

tel état de fébrilité que Zétide finit par s’en émouvoir. (FY, Grimbert La
malerune 1: Les armes des garamonts, 1998) (The sire Martin became more
and more nervous. … He was in such a state of feverishness that, in the
end, Zétide was annoyed.)
In contrast to the French findings, noun collocates in the English sub-

corpora were only significant in one genre (see Table 8): general fiction.
The only significant noun collocate in the English sub-corpora is
shock (see examples (11a/b); but see also below, Sect. 4).
(11a) Most days we found her in a state of prostration and sleepy shock, like
a survivor who has been pulled out of the rubble a week after an earth-
quake. (GEN, Banville Shroud, 2003)
(11b) Anger would come later, but for the moment he was in a state of
shock. (GEN, McEwan Solar, 2010)
2.2.3 Adjective Collocates of dans un état de

The incidence of statistically significant adjectives occurring with dans

un état de NP (see Table 9) in our results confirms Blumenthal’s obser-
vation (Blumenthal 2004, 152) that the motif dans un état de NP is
often used to indicate a (highly) advanced step in an ongoing process
(see parfait and profond ). However, we also found the statistically very
significant adjective tel, which is used quite systematically to state the
264
S. Dyka et al.
Table 9 Statistically significant adjective collocates of dans un état de NP (LLR

≥10.83)
Collocate
parfait (perfect) 62.18
proche (near) 79.25 38.06 65.65 69.85
profond (deep)
tel (such) 77.49 76.07 137.99 72.96 68.38
voisin (near) 89.10
prerequisite for a subsequent state of affairs, specified in a subordinate

clause introduced by que (that) (examples (12a/b); see below, Sect. 3,
for more on this subject).
(12a) … il se sentait dans un tel état de fureur qu ’il décida de rentrer chez
lui sur-le-champ. (ROM, Bourdin Les Années passion, 2003) (… he was in
such a state of rage that he decided to return home immediately.)
(12b) L’événement l’avait laissé dans un état de prostration tel qu ’il se
sentait incapable de penser à autre chose … (SF, Curval En souvenir du
futur, 1983) (The event had left him in such a state of prostration that he
felt unable to think about anything else …)
The most striking result, however, was the presence in most of the sub-
corpora of adjectives such as proche and voisin to express the non-typicality
of the specific state of affairs that normally the noun would express (see
examples (13a/b); see below, Sect. 3, for further discussion).
(13a) Il se rappela que Noël était dans quelques jours et cette réalité le
plongea dans un état proche de la panique. (ROM, Musso Demain, 2013)
(He remembered that it would be Christmas in a few days, and this real-
ity plunged him in a state close to panic. )
(13b) La révélation de cet épouvantable péril les plongea dans un état de
prostration voisin de l’hébétude. (SF, Guieu Opération Aphrodite, 1955)
(The revelation of this horrible danger plunged them in a state of prostra-
tion close to boredom.)
Among the English sub-corpora (see Table 10), CRIM and SF do

not show any significant adjective collocates; ROM only shows two
265
Table 10 Statistically significant adjective collocates of in a state of NP (LLR

≥10.83)
Collocate
complete 59.17
constant 133.12 154.99 50.08 65.03
extreme 63.80
great 22.69 48.00
high 25.21 76.81
permanent 94.54 66.85
perpetual 90.66
such 38.86 91.54 47.76
total 90.08
adjective collocates—constant and such—that premodify state (see exam-

ples (14) and (15)).
(14) She supposed it was the fumes from the wine that kept them all in a
constant state of lust. (GEN, Bainbridge The Bottle Factory Outing, 1974)
(15) Or if he did they would be in such a state of confusion that no one …
(HIST, Unsworth Pascali’s Island, 1980)
FY, HIST and GEN show many different adjective collocates which
premodify state or the noun in the NP. The collocates complete, extreme,
great, high, perpetual, permanent and total suggest that Blumenthal’s
findings also hold for English, since these adjectives signify the end of
a state or a high degree of its development (see Blumenthal 2004, 152).
These adjectives always premodify the noun in the NP which follows in
a state of (see example (16)).
(16) I am in a state of permanent distress. (HIST, Byatt The Children’s

Book, 2009)
2.2.4 Collocate Findings Specific to English
In contrast to the French data, only two specific collocates were found in the
English corpus: and as a conjunction is a collocate of in a state of in CRIM
and some as a quantifier in GEN. The conjunction functions as a coordinator
266
S. Dyka et al.
between two noun phrases10 that follow in a state of (see example (17)) but
can also coordinate other clause elements or clauses (see example (18)).
(17) Whatever the outcome, it could not be worse than living in a state of
perpetual uncertainty and inchoate terror. (CRIM, Dibdin Medusa, 2003)
(18) Kathleen was in a state of shock and trembling uncontrollably when
Avedissian examined the wound. (CRIM, McClure The Trojan Boy, 1988)
Some always premodifies the noun following in a state of and very often
is used as a hedge (see example (19)).
(19) In contrast to the clipped and orderly vines, the garden was in a state
of some neglect, as indeed was the outside of the house. (GEN, Mayle A
Good Year, 2004)
Other “collocates” which are highly specific for English are punctuation
marks. This indicates that in a state of NP occurs frequently at the end
of clauses followed by a comma or full stop. This might stem from the
English word order (as part of the verb valency the complement and
often adverbials follow the verb; see Sect. 3).
3 The Motifs dans un état de NP and in a

state of NP in the Corpus: A Functional
Perspective
The results from analyzing the lexicograms (see Sect. 2.2) confirm the
functional sketch of dans un état de NP given in Blumenthal (2004),
but closer examination of the specific co-occurrences also reveals a range
of other functions.
First, as pointed out earlier (see Sect. 2.2.2), dans un état de NP
allows for using nouns which do not designate “states” themselves
10It seems that and in some of these coordinated noun phrases is counted twice (once with the first noun
and once with the second noun). So the results might not be exact.
267
(even in the widest sense of the term), such as dans un état de nerfs
épouvantable, but which can function metonymically. In a similar vein,
dans un état de légume also represents a figure of speech based on a rela-
tion of similarity11; while dans un état de légume is a statistically signif-
icant co-occurrence only in the contrast corpus, the NP état de légume
can also be found in the literary sub-corpora, where (in a few examples
at least), while not statistically significant, it occurs as the prepositional
complement of réduire (reduce)12:
(20) L’action narcotique … avait … diminué sa coordination motrice

en le réduisant à l’état de légume, un légume souriant. (CRIM, Giesbert
L’Immortel, 2008) (The narcotic action … had … diminished his motor
coordination by reducing him to the state of a vegetable, a smiling vegetable.)
Expressions such as dans un état de nerfs épouvantable and dans un état

de légume clearly demonstrate that sometimes état cannot be consid-
ered as a hypernym of the entity designated by the noun following the
preposition de (as for example in état de perpléxité [state of perplexity]).
On the other hand, prostration (prostration) is defined as an “[é ]tat de
très grand abattement physique et moral” (state of great physical and
moral dejection) in the Trésor de la Langue Française informatisé (sense
B.2.b; emphasis added) or as an “[é ]tat d’abattement, de faiblesse et
d’inactivité” (state of dejection, weakness and inactivity) in the Petit
Robert (sense 2; emphasis added), whereas the lexicographic definitions
of fureur (rage) do not contain this element. Nevertheless, we find dans
un état de fureur, see example (12a), repeated here for convenience as
(21a), together with (12b)/(21b):
(21a) … il se sentait dans un tel état de fureur qu’il décida de rentrer chez
lui sur-le-champ. (ROM, Bourdin Les Années passion, 2003) (… he was in
such a state of rage that he decided to return home immediately.)
11SeeOED, “vegetable”, adjective, sense 5: “life or lifestyle … resembling that of a plant, esp.
uneventful, featureless, passive, monotonous”.
12Note, however, that (réduire ) à l’état de NP ([reduce] to the state of NP) does not represent the
same motif as (rester ) dans un état de NP ([remain] in a state of NP).

268
S. Dyka et al.
(21b) L’événement l’avait laissé dans un état de prostration tel qu’il se

sentait incapable de penser à autre chose … (SF, Curval En souvenir du
futur, 1983) (The event had left him in such a state of prostration that he
felt unable to think about anything else …)
Despite the parallelism that emerges from examples 21a/b, it must be

borne in mind that fureur is a noun which currently designates an emotion
and appears with collocates such as être (be), mettre (put) or entrer (enter),
whereas prostration does not come into play—see the examples in (22):
(22a) Il ne m’écoutait pas. Il était en fureur. Un homme fier. (GEN,

Cusset Une éducation catholique, 2014) (He did not listen to me. He was
furious. A proud man.)
(22b) Cette indifférence me mettait en fureur … (GEN, Orsenna Madame
Ba, 2002) (This indifference infuriated me …)
(22c) Alors, Skirnir entra en fureur et se mit à invectiver la géante… (FY,
Brasey La Malédiction de l’anneau, 2008) (So, Skirnir became furious and
began to insult the giant…)
In the case of dans un état de fureur, thus, the motif dans un état de NP
functions as a kind of “classifier” which shifts fureur from the realm of
emotion to that of a state—a strategy resorted to regularly when this
“state” is the reason for a certain behaviour (for a similar case in point,
see Augustyn and Grossmann (2014) as well as Novakova et al. (2018)).
Another factor which appears to trigger the use of dans un état de NP
is the function of hedging, as in constructions with proche or voisin in
French and some or near13 in English (see also above, Sects. 2.2.3 and
2.2.4, examples (13a/b) and (19)) as well as with prefixed nouns of states:
(23) Parfois, cette révélation altérait sa conscience, et l’entraînait dans un

état proche de l’ivresse. (SF, Lehman Le Cycle de F.A.U.S.T 2: Les défenseurs,
1996) (Sometimes, this revelation changed his consciousness and put him
in a state near inebriation.)
13As stated before (see above, Sect. 2.2, Footnote 7) in all cases of near (near collapse, near exhaus-
tion, near mutiny) it premodifies the noun that occurs with in a state of and functions as a hedge.
269
(24) Quant à Jean Daumale, il était entré dans un état voisin de la cata-
lepsie. (SF, Curval Rut aux étoiles, 1979) (As for Jean Daumale, he had
entered a state similar to catalepsy.)
(25) … il prenait soin de maintenir son petit appartement du quartier
Montorgueil dans un état de quasi-limpidité. (CRIM, Bruckner La Maison
des anges, 2013) (… he took care to maintain his small apartment in the
Montorgueil district in a state of almost crystal clarity.)
(26) Yet when I finally awoke it was in a state of some agitation, … (GEN,
Brookner Undue Influence, 1999)
Difficulties of clear-cut categorization of a given situation can also be

expressed by a coordination of different state nouns, which is another
situation frequently observable in the literary corpus (see also above,
Sects. 2.2.1 and 2.2.4, examples (6b) and (17)):
(27) La suite se déroula dans un état de confusion et de colère, dans un

brouillard de cris et de sang. (ROM, Musso Je reviens te chercher, 2008)
(The following took place in a state of confusion and of fury, in a fog of
shouts and of blood.)
(28) By the time they got back to the pick-up, Yongli was there in a state
of anger and frustration. (CRIM, May The Firemaker, 1992)
Another previously mentioned important function (see above,

Sect. 2.2.3) is that of introducing a state of affairs representing the pre-
condition for a subsequent state of affairs specified in a subordinate clause
that is introduced by que/that (see above, examples (12a/b)/(21a/b) and
(15)). Although less frequently, we also find examples in which the cur-
rent state of affairs is itself the consequence of a previous event:
(29) Pourquoi Kevin était-il dans un tel état de panique? (FY, Robillard
Les chevaliers d’émeraude 7: L’enlèvement, 2005) (Why was Kevin in such a
state of panic?)
A further important function of dans un état de NP seemingly relates

to the possibility of adding adjectives that serve to qualify the (usu-
ally extreme) situation designated by the noun. This qualification may
270
S. Dyka et al.
reflect a character’s point of view but is simultaneously conveyed—with

adjectives instead of (for example) relative clauses—as not depending on
an individual point of view:
(30) Dès qu’elle eut refermé la porte, il eut l’impression de recommencer

à respirer. Cette femme le mettait dans un état de stress inimaginable
[~ “que l’on n’aurait pu imaginer” (which one would not have been able
to imagine) as opposed to “qu’il n’aurait pu imaginer” (that he would not
have been able to imagine)]. (CRIM, Aubert Funérarium, 2002) (Since
she had closed the door again, he had the impression of beginning to
breathe again. This woman put him in a state of unimaginable stress.)
(31) Et je dois dire que je les ai trouvés dans un état de fatigue et d’ab-
attement alarmant [~“qui aurait alarmé tout le monde” (that would have
worried everyone) as opposed to “qui m’alarmait” (that worried me)]. (SF,
Bordage Les Griots célèbres 2: Le Dragon aux plumes de sang, 2003) (And I
have to say that I found them in a state of alarming tiredness and weariness.)
(32) Cependant, cette fois-ci, il était dans un état de fureur indescriptible
[~“que l’on n’aurait pas su décrire” (that one would not have been able
to describe) as opposed to “qu’il n’aurait pas su décrire” (that he would
not have been able to describe)] et tremblait de rage … (FY, Bujor La
Prophétie des pierres, 2002) (Admittedly, this time he was in a state of inde-
scribable anger and trembled with rage …)
In considering the level of information structure, the point can also be

made that the prepositional phrase dans un état de NP/in a state of NP
appears much more frequently in the right-hand context of the sentence
(see examples (33a/b) and (34a/b)), thus making it part of the rhematic
information, than in the sentence-initial position (see examples (35a/b))
where it expresses the “framework within which the main predication
holds” (Chafe 1976, 50).14
(33a) Il se laisse choir sur un fauteuil, dans un état proche de la stupeur.

(CRIM, Brussolo La Main froide, 1995) (He slumped down in an arm-
chair, in a state of near consternation.)
14See Riegel et al. (2014, 264–66), for the distinction between “function scénique” (framework
function) and “function rhématique” (rhematic function).

271
(33b) … j’étais bien parti pour terminer ma vie comme je l’avais com-
mencée: dans la déréliction et dans la rage, dans un état de panique
haineuse encore exacerbé par la chaleur de l’été. (GEN, Houellebecq La
Possibilité d’une île, 2005) (… I was well on the way to ending my life as
I had begun it: in dereliction and rage, in a state of hateful panic, further
exacerbated by the summer heat [Houellebecq 2005, 290].)
(34a) Ralph sat in his office assessing the reports in a state of weary disbe-
lief. (SF, Hamilton The Naked God—Flight & Faith, 1999)
(34b) Lizzie arrives for our Monday evening together in a state of very
high excitement. (ROM, Mason Playing James, 2003)
(35a) O’Brien surexcité, dans un état voisin de l’hystérie, commença de
nous noyer d’un flot de paroles que Taubelman accueillit avec calme …
(GEN, Déon Un taxi mauve, 1973) (Overwrought, in a state close to hys-
teria, O’Brien started to drown us in a flood of words which Taubelman
received calmly …)
(35b) The following afternoon, in a state of high agitation, Julia set off to
the station. (ROM, James The Queen of New Beginnings, 2010)
In summary, an important common denominator of the French motif

dans un état de NP seems to be the great morphosyntactic flexibility it
allows. This ability makes it useful for performing a rather broad range
of functions on the textual level to the point of complete syntactic inde-
pendence, as in the following—admittedly extreme—example:
(36) Tous les crimes de Schaltzmann s’étaient déroulés selon un modus

operandi analogue. Agression armée à partir de sa voiture. Carabine,
arme blanche ou cocktail Molotov au napalm. Victime choisie au hasard,
à partir de ses interprétations paranoïaques du monde. Éventuellement,
prélèvement du sang et de certains organes vitaux de la victime. Pas de
kidnapping, jamais. Pas d’utilisation de liens ou d’entraves diverses. Pas
de transport des corps. Il ne restait jamais plus de deux minutes avec
chacune de ses victimes. De l’improvisation totale, dans un état de ter-
reur cauchemardesque, d’hallucination psychotique. Une structure fractale
bien déterminée, que le Schizo-Processeur pistait jusqu’aux échos les plus
infimes de sa personnalité dérangée. (CRIM, Dantec Les racines du mal,
1995; emphasis added) (All of Schaltzmann’s crimes happened according
272
S. Dyka et al.
to an analogous modus operandi. Armed attack from his car. Rifle, knife
or napalm Molotov cocktail. Randomly chosen victim, starting from his
paranoid interpretations of the world. If necessary, removal of the victim’s
blood and of certain vital organs. No abduction, never that. No ropes or
chains were used. No transportation of the bodies. He never stayed for
more than two minutes with each of his victims. Complete improvisa-
tion, in a state of nightmarish terror, of psychotic hallucination. An exact
fractal structure, which was followed by the Schizo-Processor to the tini-
est echoes of his deranged personality.)
The quoted passage in (36) presents an interesting syntactic pattern: a first,

complete sentence is followed by seven nominal expressions which describe
in more detail Schaltzmann’s “modus operandi”; the next sentence informs
the reader about Schaltzmann’s timeline, which is again described by three
phrases without verbs, among them dans un état de terreur cauchemardes-
que, d’hallucination psychotique. The use of the motif dans un état de NP
allows for maintaining the elliptical style, describing a kind of “mixed” state
(see also cauchemardesque15) and re-classifying terreur and hallucination as
“states”, which are thus presented as reasons for the improvisation totale
(which in turn is in opposition to the rather precise “modus operandi”).
4 Stylistic Analysis of Results

At first sight, the motif dans un état de NP and its English counter-
part in a state of NP might not seem to be the most obvious starting
point, on the one hand, for seeking to link a corpus-linguistic approach
and, on the other, for asking the type of questions that scholars in the
fields of literary studies and narratology typically pursue. Our study
shows, however, that corpus-linguistics can be an eye-opener, draw-
ing our attention towards linguistic mechanisms that, while they may
15In the Trésor de la Langue Française informatisé, cauchemardesque is defined as what produces “l’im-
pression d’un cauchemar” (the impression of a nightmare) (entry “cauchemard”, derivations, 3); see
also the entry “-esque”: “Suff. formateur d’adj[ectives] dér[ivés] de noms communs, de noms propres
et d’adj[ectives] et qui indiquent une ressemblance” (Suffix used to form adjectives derived from com-
mon nouns, proper nouns and adjectives and which indicate a similarity).
273
significantly affect how literary texts work, often still go unnoticed in

traditional literary and stylistic analyses. Dans un état de NP and in a
state of NP, for instance, can be shown to play a significant role in the
textual construction of literary characters.
In particular the nominal collocates identified for dans un état de NP
and in a state of NP in our corpus (see above, Sect. 2.2.2, Tables 7 and 8)
point to what likely is one of the most important roles this motif plays
from a literary perspective: the affective discursive function (see Appendix
B), given that the majority of the nominal collocates of dans un état de NP
identified for the French corpus refer to emotional or mental states. What
also strikes us is that most of the French nominal collocates are nouns
referring to negative emotional or mental states (crise, détresse, fatigue,
fébrilité, nervosité, panique, perplexité, rage, stress, stupeur, tension, tristesse).
As pointed out previously (see Sect. 2.2.2), the use of nerf implies that
this collocate also fits into the general pattern of using dans un état de NP
to express a negative emotional state. Conscience is the only nominal col-
locate referring to a mental state that appears to have a neutral meaning.
However, depending on the modifiers combined with this particular noun,
it may turn out to have a negative meaning in specific examples after all.
Even if shock is the only statistically relevant nominal collocate for
English, it at least appears to confirm the tendency we observed in the
French data. The English corpus shows a variety of other negatively con-
noted nouns following after in a state of, as the following examples illustrate:
(37) Late that night, he wakes up; he is sweating, and in a state of high
anxiety. (GEN, Bradbury Rates of Exchange, 1983)
(38) Continued observation of colourful characters took Mather fre-
quently down cellar steps and he became in the course of time a colourful
and visionary character himself, dying at last in a state of delirium in a
Jacksonville sanatorium in 1841. (HIST, Unsworth Sacred Hunger, 1992)
Even if the noun has overall positive implications, modifiers may still
make the portrayal of a character’s mind sound more negative:
(39) He had taken an early supper with Willy, who appeared to be in a

curious state of euphoria. (GEN, Murdoch The Nice and the Good, 1968)
274
S. Dyka et al.
Use of the adjective curious here suggests that the state of euphoria
might not be an entirely positive experience. Due to the lack of statis-
tically relevant nominal collocates in the English corpus, we relied on a
manual analysis of three English sub-corpora (CRIM, GEN and ROM)
to test if our hypothesis concerning the prevalent meaning associated
with the motif would hold up—and it did so, in two respects: First, in
the majority of the examples in CRIM, GEN and ROM, the motif ful-
fils an affective discursive function. More specifically, in the majority of
examples the motif does indeed express a negative emotional or men-
tal state. Hence, our second proof: although there are instances where a
character’s mental state described by the NP following dans un état de or
in a state of can be positive or at least neutral, an important function of
dans un état de NP and in a state of NP is the depiction of emotional or
mental states, with a discernible bias towards a description of negative
states.
Presenting characters in (unpleasant) emotional and mental states in
literary texts is bound to be interesting for a literary analysis for a vari-
ety of reasons. Foremost, in constructing literary characters the depic-
tion of their emotional and mental states arguably constitutes one of the
major building blocks of fictional narratives. Providing readers with an
insight into a character’s consciousness has traditionally been deemed a
strategy that is apt to invite the reader’s empathy with, or even sym-
pathy for, these fictional “paper beings”. Arguably, texts that construct
a complex impression of the characters’ psychology are more likely to
make (and keep) readers interested in the characters, their reactions and
motivations. Authors may of course draw upon a host of different strat-
egies to achieve this effect. The motif dans un état de NP/in a state of NP
turns out to be just one of these but one that can be nuanced by fill-
ing the non-obligatory slots of the motif with a wide range of different
modifiers. This makes the motif a highly versatile device for describing,
fleshing out and varying the depiction of emotional and mental states
of characters and for expressing states that could not be captured in a
single lexeme.
Moreover, the adjectival collocates in English as well as in French
(see above, Sect. 2.2.3, Tables 9 and 10) suggest that the motif describes
especially a heightened or advanced degree of a particular emotional
275
state. Thus, the basic linguistic function of the motif dans un état de NP
and in a state of NP—potentially bridging a syntagmatic gap—also ren-
ders them interesting subjects for literary studies from the aspect of the
affective discursive function. Hedging by means of proche and voisin,
which are statistically relevant for the French corpus, and near may also
contribute to rendering more flexible and variable the description of the
characters’ emotional and mental states.
Given the motif ’s pervasive link with the depiction of negative emo-
tional or mental states and, even more, its tendency to refer to height-
ened or advanced states, it is particularly likely to occur in dramatic
and suspenseful moments in the developing narrative whenever charac-
ters experience an extreme situation (more often than not in a negative
sense). Some of the verbal collocates are also of interest in this regard.
In particular, the verbal collocate plonger in general fiction as well as in
fantasy and science fiction in the French corpus supports the hypothesis
that the motif lends itself to providing information on the characters’
psychological state at turning points, in moments of crisis, and in other
extraordinary situations. The verb plonger is ideally suited for conveying
an impression of a sudden change in the character’s state of mind or
situation.
In addition to nouns referring to mental or emotional states, the
French corpus yielded up a number of nominal collocates that may refer
to the physical states of characters: coma, faiblesse, prostration, saleté (see
above, Sect. 2.2.2, Table 7). Again, the nouns clearly have negative con-
notations. If these nouns occur in the construction dans un état de NP,
the motif is likely to have a descriptive discursive function, which may
also serve to flesh out the description of a literary character indirectly
by providing information on his/her environment. Beyond that, some
of the nominal collocates also suggest that the motif dans un état de NP
may have a more broadly descriptive discursive function, potentially
serving to describe a setting. This seems to be particularly likely with
the nominal collocates saleté, manque and délabrement. In this context,
it seems interesting that délabrement is statistically relevant as a collocate
in crime fiction, where one might expect the expression dans un état de
délabrement to refer to a crime scene or even to a corpse.
276
S. Dyka et al.
5 Conclusion
Our analysis of the expression dans un état de NP and its English coun-
terpart in literary texts of different subgenres (CRIM, FY, GEN, HIST,
ROM, SF) confirmed Blumenthal’s observations concerning its use in
French newspaper texts. But, it also revealed a broader range of func-
tions that seem to be specific to literary texts and thus justify catego-
rizing dans un état de NP/in a state of NP as a motif. The differences we
found in the usage of this motif—including that it appears to be used
more restrictively in certain subgenres—concern both the linguistic
level and the level of the different subgenres. French and English seem
to employ this expression differently, ranging from the absence of sta-
tistically significant verbs denoting “change of state” in the English cor-
pus to the systematic statistical significance of be in the English corpus
whereas its equivalent être is absent in the French corpus. As for the dif-
ferences between subgenres, undoubtedly further investigation is needed
in order to identify what is specific to a certain subgenre written in one
language as opposed to its counterpart in the other language. Of rele-
vance here will be contrasting adjectives in English fantasy novels with
their French counterparts. We also found evidence of tendencies that
both languages share: the motif is used with more verbs, adjectives and
nouns in GEN than in (almost) any of the other subgenres (see Table 9
for adjectives in French SF). This suggests that GEN indeed shows (a set
of ) specific features which require still more detailed investigation using
corpus-linguistic tools.
More generally speaking, since the psychology of fictional characters
tends to play a vital role in literary texts captivating readers in the first
place, the linguistic and stylistic strategies used to construct these “paper
beings” clearly deserve more attention from literary studies. The present
study highlights the practical possibilities offered by corpus-linguistic
approaches in this context. As we have shown, the construction dans
un état de NP/in a state of NP contributes a certain variety to the pres-
entation of a character’s psychology—highly desirable for writers of fic-
tion. Thus, the functions of the motif dans un état de NP/in a state of
NP which we identified beyond those discussed by Blumenthal (2004)
appear to be driven by stylistic considerations.
277
While our analysis has demonstrated a notable difference in the

textual functions of the motif dans un état de NP/in a state of NP
between literary and non-literary texts (see Blumenthal’s results versus
those obtained from the contrast corpus), we could not do more than
add another piece to the complex jigsaw puzzle of literary subgenres.
Nevertheless, taken together with the other motifs discussed in this vol-
ume, which have turned out to be equally useful categories for compar-
ing (sub)genres, it makes a contribution to the larger picture gradually
starting to take shape.
References
Augustyn, Magdalena, and Francis Grossmann. 2014. “Entre hyperonymie
et spécification: un drôle de sentiment.” In Les émotions dans le discours –
Emotions in discourse, edited by Peter Blumenthal, Iva Novakova, and Dirk
Siepmann, 123–34. Frankfurt am Main: Peter Lang.
Blumenthal, Peter. 2004. “Definition und Wortgebrauch: Zustandsnomina im
Französischen.” In Romanische Sprachwissenschaft: Zeugnisse für Vielfalt und
Profil eines Faches. Festschrift für Christian Schmitt zum 60. Geburtstag, vol.
2, edited by Alberto Gil, Dietmar Osthus, and Claudia Polzin-Haumann,
143–78. Frankfurt am Main: Peter Lang.
Chafe, Wallace. 1976. “Givenness, Contrastiveness, Definiteness, Subjects,
Topics, and Point of View.” In Subject and Topic, edited by Charles Li,
25–55. New York, San Francisco and London: Academic Press.
Houellebecq, Michel. 2005. The Possibility of an Island. Translated by Gavin
Bowd. London: Weidenfeld & Nicolson.
Houellebecq, Michel. 2016. Submission. Translated by Lorin Stein. London:
Vintage.
Kraif, Olivier, and Sascha Diwersy. 2012. “Le Lexicoscope: un outil pour l’étude
de profils combinatoires et l’extraction de constructions lexico-syntaxiques.”
In Actes de la conférence conjointe JEP-TALN-RECITAL 2012, Volume 2:
TALN, edited by Georges Antoniadis, Hervé Blanchon, and Gilles Sérasset,
399–406. Grenoble: Association Francophone pour la Communication
Parlée and Association pour le Traitement Automatique des Langues. http://
www.aclweb.org/anthology/F12-2033.
278
S. Dyka et al.
Kraif, Olivier, and Sascha Diwersy. 2014. “Exploring Combinatorial Profiles

Using Lexicograms on a Parsed Corpus: A Case Study in the Lexical Field
of Emotions.” In Les émotions dans le discours – Emotions in discourse, edited
by Peter Blumenthal, Iva Novakova and Dirk Siepmann, 381–94. Frankfurt
am Main: Peter Lang.
Le Petit Robert de la langue française. 2019. Version numérique 5 (2). http://
pr12.bvdep.com.
Lexicoscope. http://phraseotext.univ-grenoble-alpes.fr/lexicoscope.
Longrée, Dominique, and Sylvie Mellet. 2013. “Le motif: une unité
Boston: De Gruyter Mouton.
Novakova, Iva, Julie Sorba, and Agnès Tutin. 2018. “Les noms généraux senti-
ment et émotion.” Langue française 198: 83–97.
Oxford English Dictionary Online. Oxford University Press. http://www.oed.
com.
Riegel, Martin, Jean-Christophe Pellat, and René Rioul. 2014. Grammaire
méthodique du français, 5th ed. Paris: Presses Universitaires de France.
Teich, Elke. 1999. Systemic Functional Grammar in Natural Language
Generation: Linguistic Description and Computational Representation.
London and New York: Cassell.
Trésor de la Langue Française informatisé. http://www.atilf.fr/tlfi.
Vargas, Fred. 2014. Dog Will Have His Day. Translated from the French by
Siân Reynolds. London: Harville Secker.
10
Towards an Interdisciplinary Approach
for Differentiating Contemporary
Fiction Subgenres
To conclude our exploratory journey through contemporary French,

English and American literature using modern lexico-statistical m
ethods
and an exhaustive corpus-driven approach, we summarize here our
main results and the perspectives on productive future research in
digital humanities they have opened. All the contributions in this

volume converge into three major avenues of research that examine:
(1) the importance of Natural Language Processing (NLP) and the new

methodology we developed in the PhraseoRom project for a lin-
guistic and stylistic analysis of contemporary literature;
I. Novakova (*)
D. Siepmann
https://doi.org/10.1007/978-3-030-23744-8_10
280
(2) the viability of using statistically significant expressions and textual

motifs in identifying literary subgenres; and,
(3) the comparison of these expressions and textual motifs in English
and French literary subgenres.
Let us briefly recap the stages of the journey that brought us to this
point.
We started out by discussing previous research into phraseology, sty-
listics and the theories of literary genre, each of which makes a specific
contribution to describing recurrent lexico-grammatical patterns in con-
temporary fiction. We also presented our newly designed methodology
based on lexical statistics approaches to extracting recurrent lexico-syn-
tactic patterns (Chapter 1) and then moved on to an overview of the
term “motif ” in different disciplines (Chapter 2). Despite the apparent
diversity in the use of this “crossroads” term, it has emerged that the
concept of motif essentially refers to recurrent patterns in folkloristics,
narratology, bioinformatics, NLP and linguistics. Pivotal in our attempt
to characterize such recurrent patterns (i.e., our recurrent lexico-syn-
tactic trees or RLTs) was Longrée’s and Mellet’s definition of the term
(2013, 2018), which holds that motifs are “an ordered subset of the
textual ensemble, formed by the recurring combination of n elements”
(Longrée and Mellet 2018, 156) and which fulfil specific, identifiable
functions on the semantic, pragmatic and rhetorical levels (see Longrée
and Mellet 2013, 66).
Taking two different approaches (keywords analysis and correspond-
ence analysis), the next two Chapters 3 and 4 converge in their finding
that the writing style is more variable and creative in French novels than
in British or American novels. For example, while—sometimes rather
banal—adverbs ending in -ly (to walk briskly) are very frequently used
in British and American novels, manner adverbs ending in -ment tend
to be avoided in French novels and are replaced by various periphra-
ses, as in this case, by marcher (walk)/s’éloigner (move away)/se diriger
(go towards)/se lever (stand up)/traverser (get out)/monter (climb)+ avec
raideur (with stiffness)/d’un pas raide (with a stiff step) (see Chapter 3
by Novakova et al. in this volume). Our results show that in French
the motifs generated by manner adverbs are more complex, furnishing
10 Towards an Interdisciplinary Approach for Differentiating …
281
empirical proof of French literary language’s higher variability. French

writers also embellish these structures in ways that English is unlikely
to adopt. Thus, we found motifs like se lever avec une lenteur majestueuse
(rise with majestic slowness)/avec une grâce troublante (rise with disturb-
ing grace) or hocher la tête avec une brièveté calculée (nod with calculated
brevity). While such expressions are systemically possible in English—
and indeed were used in the eighteenth and nineteenth centuries—they
are probably rather noun-heavy for modern tastes and thus tend to be
replaced by more succinct constructions, e.g. nod briefly/stiffly in this
instance. To put it in colloquial terms, we might say that French authors
“are into” the subtler shades of meaning that such prepositional phrases
can pack in lieu of adverbs.
Another striking finding the book owes to Chapter 4 is how pro-
nounced a role the generic verb to say plays in English corpus speech
tags—it accounts for no less than 63% of verba dicendi in the English
data set. The pervasiveness of this semantically unmarked or “neutral”
verb suggests that lexical variation in the speech tag construction’s
verb slot is significantly lower in English than in French novels. The
“dogma of avoiding repetition” of the same word (Philippe 2016, 69)
has undoubtedly played an important role in heightening the creativ-
ity of French writers since the nineteenth century. For English novel-
ists, on the other hand, this lexical variation so prevalent an ideal in the
French tradition is not a prime stylistic consideration. Our results sug-
gest, therefore, that the distributional configurations of speech verbs in
English and French fiction as well as in the different genres (crime, fan-
tasy and romance) can be accounted for in part by stylistic preferences
of the two languages. Also, the predilection for generic verbs like say can
be interpreted as trying to hold narrator intervention to a minimum,
to foster the illusion of the narrator not filtering the characters’ utter-
ances. The high frequency of expressive speech verbs in French popular
fiction (fantasy, romance, crime) suggests a strong tendency to specify
the speaker’s attitude. Conversely, several scholars have observed that
English fantasy tends to favour a comparatively simple language.
English and French fiction also differ in how each depicts alcohol
and tobacco consumption, as observed in Chapter 5. French writers,
for instance, are more likely to use the generic term “alcohol”, while
282
English authors seem more disposed to using specific references (to

different types of alcohol). In French crime fiction and romance novels,
references to alcohol typically serve to portray addiction or personal
decay or evoke a medical context. This chapter delved into scripts
related to alcohol and tobacco consumption and the motifs generated
by the expressions light a cigarette/allumer une cigarette and have a
drink/boire un verre in English and French. It revealed that the expres-
sion take a sip is characteristic of romance novels in English whereas
its French near-equivalent boire une gorgée is not. Despite alcohol and
tobacco consumption having been a staple in literary texts since the
1950s, as shown by our corpus-driven approach, it has so far not been
examined systematically—either from a literary or linguistic perspective.
Smoking a cigarette or drinking an alcoholic beverage is mainly used,
across all genres, for representing common forms of human interaction.
Having characters imbibe or light up is also a device for announcing
or suspending dialogue or action. Therefore, the motifs in this chapter
were systematically related to their—descriptive, affective, cognitive or
infra-narrative—discursive functions (see also Appendix B).
Chapters 6 and 7 moved on to considering the role that “fiction
words” and collocations play in distinguishing literary subgenres.
Focusing on two comparable sets of randomly drawn fiction words, in
Chapter 6 the authors examined how these words are created morpho-
logically as well as the semantic fields they belong to in 1990s American
and French science fiction. Using data obtained with the RLT extrac-
tion technique, the authors of this chapter compared their syntactic dis-
tributions. The presence of “transfuges” or fiction words—neologisms
such as holo, psionic, terraform, astrogation, conapt and cyborg that leap
not only across the boundaries of works by individual authors but also
hurdle language barriers—revealed a great unity in this genre-specific
imaginary landscape where inventions and findings abound. A compa
rison between an American and a French science-fiction corpus here
also showed that the two different literary traditions may share a com-
mon background of fictional references. Although French SF is part of a
specific French literary tradition dating back to influential authors such
as Jules Verne, the separate science fiction universes created by works
in American and French literary cultures in effect ended up merging in
283
recent decades. This fusion was perhaps inevitable, given the outsized
influence epitomized by the huge, unreciprocated one-way flow of
translations from (American) English into French.
Chapter 7 applied a semantic categorization of lexico-syntactic pat-
terns (RLTs) to distinguish between two closely related narrative sub-
genres, science fiction (SF) and fantasy (FY). The RLTs extracted from
the two corpora were classified in an exhaustive semantic grid. It served,
for example, for classifying the recurrent lexico-syntactic trees specific
to the SF corpus, such as s’enfoncer | sortir | pénétrer dans la forêt | les
bois (to go deep into | to emerge from | to enter the forest | the woods),
in the <action: travel> semantic category. These RLTs generate different
motifs related to a journey through a specific landscape, which could be
a forest, a wood, a castle or a battlefield. Literary critics do not distin-
guish between science fiction and fantasy based on linguistic differences,
but instead on an opposition of the imagined and the reader’s experi-
ence. Our methodology can highlight these “generic markers” (the
medievalism of fantasy, for example, or the salience of specific spaces
(space, forest) and certain “stereotyped actions” like the martial dimen-
sion of characters in fantasy novels). The results for both the French
and English corpora showed that the boundary between the two gen-
res is generally respected in them. However, a more detailed analysis of
certain categories (such as cognition and, especially, communication)
revealed specificities in each language. This grid, developed for the pur-
poses of PhraseoRom, represents one of our project’s key achievements.
After analyzing microstructures (fiction words or collocations) in the
two previous chapters, the focus in Chapters 8 and 9 shifts to specific
motifs defined as characterizing and structuring elements (Longrée and
Mellet 2013) within the fictional world. Chapter 8 covers the motifs of
writing écrire/lire_DET_lettre/roman, write/ and reading read_DET_let-
ter/_novel specific to general fiction (GEN) as opposed to crime novels
(CRIM). These structures revealed that characters in GEN are depicted
reading more letters and books than those in CRIM (and we would
probably find the same situation in other subgenres). General litera-
ture is replete with this type of action and personal exchanges between
the characters. The results also revealed that écrire_DET_roman/write_
DET_novel is most characteristic of French general fiction, leading to
284
the hypothesis that the French literary novel is more metafictional than
the English one. The recurrent lexico-syntactic constructions found in
our corpora show significant syntagmatic and paradigmatic variations.
They also perform various discursive functions which may be narrative
and descriptive but can also be cognitive, pragmatic or commentary.
More generally, we regard the identification of specific motifs and their
discursive functions as potentially contributing to the elaboration of “an
operating theory of genres” (Rastier 2011, 72).
In Chapter 9, other interesting motifs generated by the collocations
“dans un état de NP/in a state of NP” undergo analysis. Comparing the
two languages showed that in a state of NP is more frequent in English
corpora than dans un état de NP in French corpora in all subgenres
except science fiction. The results for crime novels demonstrated a clus-
tering with nouns related to nervousness: tension (tension), nervosité
(nervousness), nerfs (nerves), while science fiction novels instead tend
to be characterized by depictions of physical exhaustion: faiblesse (weak-
ness), prostration (prostration) and fatigue (tiredness). Fantasy fiction,
in turn, favours nouns indicating a kind of helplessness dans un état de
fébrilité (in a state of anxiety). Motifs like dans un état de nerfs épouvant-
able (in a dreadfully nervous condition), dans un état de fureur indescrip-
tible (in a state of indescribable fury) indicate how characters’ emotional
states are depicted. Another interesting result related to general fiction
revealed that the motifs here are used with more verbs, adjectives and
nouns than in any other genre, possibly indicating that in general fic-
tion they are endowed with less specific features than in other genres.
To sum up, we undertook our exploration in this volume to demon-
strate the heuristic power the notion of motif—a notion not well-
known in Anglo-Saxon phraseology—holds for linguistic and stylistic
analysis of literary texts. Our methodology has contributed to advan
cing how we think about the textual motif as a structural element of
literary texts and as a distinctive marker of (the) fiction genre(s). This
idea has proven to be especially fruitful for our research since it allows
us to relate the micro-level of recurring multi-word units to the macro-
level of fictional scripts. We hope that the journey detailed in this book
will help guide future research towards refining a functional typol-
ogy of motifs and formalizing the distinction between literary fiction
285
and popular fiction (given that authors of literary fiction, for example,
gleefully rework motifs supplied by popular fiction, as for example by
replacing “there was silence” with “an extraordinary, ringing silence
descended” [John Banville]). Finally, we are confident that our corpus-
linguistic approach based on the identification of recurring linguistic
patterns can provide both the tools and the evidence needed for a better
understanding of fictional scripts. This approach thus may prove to be
an important contribution to cognitive narratology.
References
Boston: De Gruyter Mouton.
Philippe, Gilles. 2016. French Style. L’Accent français de la prose anglaise.
Brussels: Les Impressions Nouvelles.
Rastier, François. 2011. La Mesure et le grain. Sémantique de corpus. Paris:
Honoré Champion.
Appendix A: Corpora
See Tables 1, 2, 3, and 4.
© The Editor(s) (if applicable) and The Author(s) 2020 287

https://doi.org/10.1007/978-3-030-23744-8
Table 1 Size of the entire comparable corpora
French English German
288 Appendix A: Corpora
Aut.a Nov. Tokens Aut. Nov. Tokens Aut. Nov. Tokens

FY 43 104 13,323,976 41 142 17,814,599 40 109 15,800,178
GEN 170 445 34,334,554 89 147 17,442,917 41 90 8,413,707
HIST 39 114 14,868,273 48 116 17,764,261 38 149 22,932,408
CRIM 84 194 17,859,351 55 115 13,561,947 34 97 10,415,128
ROM 40 112 9,802,410 35 96 13,295,474 46 208 17,715,657
SF 39 147 13,173,618 39 116 12,962,491 38 91 8,094,857
Total 365 1,116 103,362,182 274 732 92,841,689 201 744 83,371,935
aAut = author, Nov = novel
Table 2 Size of the samples in the literary corpora (LIT) (Samples of literary corpora (LIT) versus reference corpora
[CONT] [cf. Table 3])
LIT (French) LIT (English) LIT (German)
Aut. Nov. Tokens Aut. Nov. Tokens Aut. Nov. Tokens
FY 35 38 4,937,890 36 36 5,149,855 34 34 5,047,280
GEN 79 79 5,021,345 46 46 4,902,324 41 53 5,092,698
HIST 38 47 5,186,613 35 35 5,010,151 36 36 5,063,660
CRIM 52 56 5,355,426 46 46 5,071,283 34 50 5,075,617
ROM 40 58 4,535,997 34 39 4,961,251 46 56 4,996,205
SF 38 58 4,963,265 36 39 5,117,483 38 47 4,948,965
TOTAL 253 336 30,000,536 214 241 29,212,347 162 276 30,224,425
Appendix A: Corpora 289
290 Appendix A: Corpora
Table 3 Size of the contrast (non-literary) corpora (CONT)

Tokens CONT (French) CONT (English)
TV subtitles 9,867,437 21,867,400
TED talks 3,225,622 –
Debates (Parliament) 12,695,516 15,106,386
Science papers 14,392,200 17,821,043
News 4,579,020 4,062,972
Press 20,283,232 20,183,187
TOTAL 65,043,027 79,040,988
Table 4 Size of the parallel corpora

Source language: French Source language: English
Aut. Nov. Tokens FR Tokens EN Aut. Nov. Tokens EN Tokens FR
FY 3 7 1,058,007 840,055 48 139 20,986,290 20,568,205
GEN 45 68 5,786,001 5,977,204 56 75 8,775,209 10,056,224
HIST 9 18 2,924,227 2,276,972 38 66 11,591,960 10,843,217
CRIM 15 25 2,311,310 2,353,667 30 49 5,296,693 5,669,642
ROM 6 7 494,371 561,120 17 24 3,407,815 3,548,149
SF 5 5 539,868 527,136 33 69 8,486,630 8,284,597
OTHER 5 5 488,030 480,543 36 71 8,453,438 8,978,998
TOTAL 88 135 13,601,814 13,016,697 258 493 66,998,035 67,949,032
Appendix B: Discursive Functions
Synthesis on Discursive Functions
Definition
The discursive function (DF) of a motif depends on the role it plays in

the context, such as in maintaining “textual coherence” and “thematic
continuity” (Martin 1983, 15–16). In the PhraseoRom project, we
understand “discursive” as “related to fiction discourse”.
The same motif may have different DFs, again depending on its con-
text. Thus, merely identifying the motif is not the same as analysing it
stylistically: it must still be interpreted in the sentence that contains it
or, even more broadly, by contextualizing its uses (see Adam 2011, 50;
Baroni 2015, 140).
Method
Consistent with the corpus-driven approach, we followed an induc-

tive method. In pilot studies conducted in recent years on the French

https://doi.org/10.1007/978-3-030-23744-8
292 Appendix B: Discursive Functions
corpora, we worked on several motifs. Based on the RLTs provided and

grouped by the computer program, here we identify as motifs the LSCs
that demonstrate syntagmatic (right and left contexts) and paradigmatic
variations (several instances of a single motif, as in | immediately | rag-
ingly | lighting | crushing | a | his | new menthol | cigarette | gitane | against
a wall |) and that also perform discursive functions in the text. These
functions are often codified as playing the same routine roles in novels.
Stylisticians determine the different discursive functions a motif
performs by observing how it occurs in several different contexts. The
list of discursive functions below is therefore culled from a variety of
sources. It is a work in progress and as such will likely need to be devel-
oped further. The examples, extracts from French corpora that we have
translated into English, are merely illustrative here. The motifs in the
examples are italicized.
Discursive Functions
(i) Narrative and descriptive DFs are encountered mostly in novels

(Adam 2011, 267).
(1) “the driver consulted his watch: 8:15 a.m.” (Narrative DF,
with the motif playing an active role in the plot.)
(2) “He looked out the window again. Cooking colours, that’s
what the colours of Italy were…” (then follows a descrip-
tive sequence, triggered by the motif ).
(ii) Affective DF: a motif referring to affects represents a special case
of the descriptive function.
(3) “Sarah nervously crushed her cigarette.”
(iii) Indirectly descriptive DF: a repeated action, a typical gesture
describing the character. For instance, a heavy smoker in CRIM
novels; with increased awareness in FY.
(iv) Infra-narrative DF: refers to DFs in the background of the
action. The motifs serve to embellish the conversation in a
script, without narrative consequences for the main action.
(4) “‘You’ll do better next time,’ Alexandre asserts, lighting a
cigarette.”
Appendix B: Discursive Functions 293
(v) Infra-descriptive DF: the motif provides a minimal, often stere-

otypical descriptive precision.
(5) “Now they were silent, looking out the window at the
reflections of a sinister sky in the waters of the lagoon.”
(vi) Cognitive DF: these motifs apply to cognitive processes
(hypotheses, apprehension of events, reflections, etc.).
(6) “‘I don’t know what’s going to happen to him. I can’t afford
to change him from one school to another.’”
(vi 1) Remembrance DF: this is a special case of the cognitive func-
tion used to relate cognition to the expression of memories.
(vi 2) Commentary DF: another special use of the cognitive function
that relates cognition to a reflection on writing activity (found
only in the GEN FR corpus).
(7) “‘Of course, one could have considered writing a jet set
Proustian novel …; it would have had no interest.’”
(vii) Pragmatic DF: applies to motifs that express speech acts
between the characters in the novel (mainly direct speech).
These motifs establish coherent relationships between characters
within the stated discourse integrated into the narrative text.
(8) “‘Do not do anything about it, Madam, I beg you,’
Eudeline cried.”
References
Martin, Robert. 1983. Pour une logique du sens. Paris: PUF.
Adam, Jean-Michel. (1992) 2011. Les Textes: types et prototypes, 3rd ed. Paris:
Armand Colin.
Baroni, Raphaël. 2015. “Temps, mode et intrigue: de la forme verbale à la
fonction narrative.” Modèles linguistiques, 71: 125–142.
Index
A Colligations 4, 167, 169

Aarne-Thompson classification 20 Collocates 55, 56, 61, 65, 67, 225,
Alcohol and tobacco consumption 229–231, 254–260, 262–266,
vii, 115, 116, 124, 281, 282 268, 273–275
Allomotif 24 Collocational “nests” 4
Argument schema 39 Combinatory 225
Computational linguistics vi, 18
Construction grammar 5, 39, 40, 42
B Corpus analysis 98
Bioinformatics vii, 18, 26, 280 Corpus and dictionary study 89
Brightly 53, 56, 64, 65, 68, 72, 73 Corpus linguistics 9, 48, 87, 93, 99,
Briskly 50, 53, 55–57, 60, 61, 70, 123, 153, 154, 272, 276, 285
71, 280 Corpus stylistics vii, 9, 49, 87
British contextualism 5 Correspondence analysis 9, 93, 94, 280
Creative writing 11, 79
Creativity 79, 118, 204, 281
C Creativity principle 3
Clause 10, 23, 25, 30, 173, 182, Crime fiction viii, 6–8, 35, 37, 39,
237, 241, 242, 264, 266, 269 43, 86–89, 93, 95, 98, 101,
Collective memory 20 108, 109, 120, 123, 124, 130,

https://doi.org/10.1007/978-3-030-23744-8
296 Index
132, 135, 142, 144, 205, 215, Fiction words vii, 105, 152–160,
218, 223, 228, 247, 256, 261, 162, 165, 167, 168, 170,
275, 281–284 172–178, 180–186, 194, 282,
283
Functional equivalents vii, 47
D
Digital humanities vi, viii, 279
Digital stylistics 6, 7, 79 G
Dimly 53, 56, 64, 65, 67, 72 General fiction vi, 3, 8, 102, 104,
Direct speech vii, 52, 84–89, 99, 110, 124, 132, 145, 256, 258,
100, 102–104, 106, 110, 182, 263, 275, 283, 284
215, 239, 242, 243, 293 Genericity 6
Discursive function vii, 37, 42, 54, 73, Generic markers 204, 207, 208, 210,
76, 120, 121, 126, 128, 129, 213, 219, 283
131, 134, 147, 200, 219, 226, Grimm brothers 20
227, 229, 235–241, 243–247,
273–275, 282, 284, 291, 292
H
Highbrow literature 6, 7
E Historical novel vii, viii, 110, 218,
Extended lexical units 5 256
Extended phraseology 5 Hybridization of genres vi, 190
F I
Fabula 23 Idiomaticity vii, 1, 3, 4
Faintly 53, 56, 64–66, 72 Idiom principle 3
Fantasy viii, 6, 7, 8, 35, 70, Index of Motifs 20
84, 86–89, 93, 95, 96, 98, Intertextuality 20
101–109, 123, 124, 145, 146,
190–194, 197, 200–216,
218–220, 256, 263, 275, 276, K
281, 283, 284 Key adverbs vii, 51, 78
Fictional genres v–vii, 87, 110, 152 Key manner adverb 54, 78
Fictional scripts 10, 284, 285 Keyword 8, 48, 49
Fiction-specific adverbial construc- Keyword approach 7–9, 35, 93,
tions 8 280
Fiction-specific keywords 4 Key word patterns 38
Index 297
L Motifs of Motion and Light 60

Lexical bundles 4, 37 Motif texture 25
Lexicoscope 8, 35, 154, 227, 229,
230, 254, 255
Literariness vi, 2, 3 N
Literary fiction 6, 205, 218, 223, Narrative function 10, 24, 84, 128,
227, 235, 242, 247, 284 129, 240, 245
Literary language 2, 3, 9, 47, 49, 51, Narratology vii, 10, 18, 22, 23, 25,
79, 109, 151, 281 26, 42, 123, 272, 280
Literary translation vi cognitive narratology 285
Log likelihood 9, 49 Natural equivalents 50
Log-likelihood ratio (LLR) 35, 227, 255 Natural Language Processing (NLP)
Lowbrow fiction 6, 7 vi, 18, 279
Neologisms vii, 151–153, 155–157,
166, 194, 282
M Nervously 50, 51, 53–55, 73–78,
Manner adverbs 49, 51, 53–56, 79, 131, 292
280 N-grams 36, 37, 55
Märchentypen 20
Minimal thematic unit 22
Minimum unit 23 O
Motif detection 27 Occasionalisms 156
Motifemes 24 Open-choice principle 3
Motifs 1, 5–10, 17–43, 47, 51, 55,
57, 60, 61, 63, 64, 66, 68, 69,
75, 77, 79, 116, 129, 136, P
143, 146, 151, 152, 200, 205– Phraseological unit 1, 2, 5, 39,
208, 210–214, 219, 224–227, 223–225, 227, 229
229, 231, 235–247, 254, 256, Phraseology v–vii, 1, 5, 152, 154,
257, 263, 267, 268, 271–277, 223, 225, 280, 284
280–285, 291–293 PhraseoRom v, 7, 19, 35, 37, 49,
dynamic 23 54, 88, 110, 124, 151, 152,
F-motif 31 154, 155, 189, 191, 193, 195,
sequential motifs 33 216–220, 246, 247, 254, 279,
stative 23 283, 291
valency motifs 31 Popular genres 6, 102
world length motifs 29
298 Index
R Subgenres viii, 7, 8, 35, 87, 89, 151,

Recurrent fiction-specific patterns v 153, 154, 184, 189–193, 197,
Recurrent lexico-grammatical pat- 214, 215, 217–219, 239, 253,
terns vii, 1, 280 256, 257, 259, 260, 276, 277,
Recurrent lexico-syntactic trees 280, 282–284
(RLTs) 8, 10, 35–37, 124, 125, Syngrams 38, 47
128, 129, 131, 134, 136, 139, Syntactic dependency trees 35
144, 148, 154, 155, 167, 169, Syntagmatic and paradigmatic varia-
189, 191, 193–195, 198, 200, tions 229, 231, 235, 247, 284
203–216, 280, 283, 292
Romance viii, 6, 7, 8, 20, 86–89, 93,
98, 101, 102, 105, 108, 109, T
124, 142–146, 148, 215, 218, Taletypes 20
281, 282 Textometry 2, 9, 34, 93, 153
Russian Formalists 23 Textual colligations 4
Theories of literary genre vii, 1, 280
Theory of genres 6, 227, 284
S Translation equivalents 50, 52
Science fiction viii, 7, 8, 35, 70,
84, 110, 123, 124, 145, 149,
151–153, 159, 160, 174, V
184–186, 190–194, 200, 201, Verba dicendi. See Speech verbs
203–216, 218–220, 256, 261, Verb of motion 60
275, 282–284 Verbs denoting light 64
Semantic categorization 191, 200,
211, 216, 283
Semantic grid 160, 194, 200, 202, X
203, 205, 216–220, 283 Xenoencyclopedia 152, 154, 166,
Sketchengine 8 176, 187, 207
Social scripts 10
Speech verbs 9, 52, 70, 71, 86–89,
92, 95–100, 102, 104–106, Z
108–110, 215, 281 Zipf-Mandelbrot distribution 31
Stiffly 53, 56, 60–63, 71, 72, 281 Zipf-Mandelbrot type 30
Story/narrative distinction 23
Stylistics vi, vii, 1, 2, 6, 9, 11, 31,
52, 79, 124, 191, 280

Phraseology and Style in Subgenres of The Novel

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Phraseology and Style in Subgenres of The Novel

Uploaded by

Copyright:

Available Formats

Phraseology and

ISBN 978-3-030-23743-1 ISBN 978-3-030-23744-8 (eBook)

© The Editor(s) (if applicable) and The Author(s) 2020

Cover credit: Marina Lohrbach_shutterstock.com

The genesis of this book was in a four-year collaborative research project

1https://phraseorom.univ-grenoble-alpes.fr. The PhraseoRom project (2016–2020), led by

3Deutsche Forschungsgemeinschaft (DFG).

compares and contrasts the stylistic practices encountered in British,

Presentation of the Chapters

Chapter 1, written by I. Novakova and D. Siepmann, outlines lin-

Chapter 9 (S. Dyka, L. Fesenmeier and M. Gymnich) studies the

Grenoble, France Iva Novakova

We owe a great debt of gratitude to Professor Marion Gymnich

1 Literary Style, Corpus Stylistic, and Lexico-Grammatical

2 The Notion of Motif Where Disciplines Intersect:

3 Key Adverbs and Adverbial Motifs in English Fiction

4 Speech Verbs in French and English Novels 83

5 Alcohol and Tobacco Consumption in English

6 French and American Science Fiction During

7 Science Fiction versus Fantasy: A Semantic

8 Reading and Writing as Motifs in English

9 Dans un état de NP and in a state of NP: Bridging

10 Towards an Interdisciplinary Approach for

Appendix A: Corpora 287

Appendix B: Discursive Functions 291

Sascha Diwersy is associate professor of French linguistics in the

Erlangen-Nuremberg, Germany. His main research interests are in

Clémence Jacquot is an associate professor of linguistics and stylis-

French-German Emolex (2010–2013) research project EMOLEX on

CONT Contrast corpus

Fig. 2 “Other” types of alcohol in the English and French

Table 3 Semantic dimensions and values used

significantly, this assumption of a functional difference without a formal

to linguistic features and their functions but in which they failed to

2.2 Approaches to Idiomaticity and Phraseology

As we have just seen, while there is a dearth of studies on the lexi-

Construction grammars, on the other hand, treat language as an

2.3 Stylistics and Theories of Literary Genre

Textual genres are currently the subject of interdisciplinary studies in

3 Corpora and Methodology

novels, subdivided into the six subgenres of crime, science fiction,

– the RLTs5 (Recurrent Lexico-syntactic Trees) method, integrated into

The RLTs methodology automatically extracts statistically significant

(Arbres Lexico-syntaxiques récurrents; Kraif 2016).

is by now a well-established concept in corpus linguistics: A word must

4 Innovative Features of the Book

– display lexico-syntactic regularities and variations at the syntagmatic

units that can be described at the levels of lexico-grammar, semantics

Motifs as we understand them cannot be identified by fully auto-

The present volume is intended for linguists, translation and literary

Scholes, Robert. 1982. Semiotics and Interpretation. New Haven: Yale

This success can be gauged by the proposals and innovations advanced

PhraseoRom project on which all of the contributions in this book are

2 Motif in Folkloristics and Narratology

2.1 Tales and Popular Narratives

For our purpose, the study of motifs in tales or popular narratives is of

(1) - “Punishment, sitting in the water”1, which appears in many novels

The conception of motif in ancient and traditional texts necessarily

Thompson divided each category into subcategories. For example, the

- J0-J199 Acquisition and possession of wisdom/knowledge

A motif is the smallest element in a tale having a power to persist in tra-

A current conception of motif in narratology seems to echo Thompson’s

Motif: A discrete thing, image, or phrase that is repeated in a narrative.

CONT Contrast corpus

2.2 Approaches to Idiomaticity and Phraseology

2.3 Stylistics and Theories of Literary Genre

3 Corpora and Methodology

4 Innovative Features of the Book

2 Motif in Folkloristics and Narratology

2.1 Tales and Popular Narratives

2.2 The Russian Formalists

2.3 Some Developments of the Formalist Approach

3 From Narratology to Bioinformatics

4 Motifs in Automatic Text Processing

5 The Contribution of Computer Linguistics

5.1 The Ganascia (2001) Approach

5.2 The Approach of Longrée and Mellet

5.3 The RLTs Approach

5.4 Motifs and Constructions