in Coherence of Discourse Formal and Conceptual Issues of Language Maxime Amblard Online Ebook Texxtbook Full Chapter PDF

In coherence of Discourse Formal and
Conceptual Issues of Language Maxime

Amblard
Visit to download the full and correct content document:
https://ebookmeta.com/product/in-coherence-of-discourse-formal-and-conceptual-issu
es-of-language-maxime-amblard/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Language and Law in Professional Discourse Issues and

Perspectives 1st Edition Vijay K. Bhatia
https://ebookmeta.com/product/language-and-law-in-professional-
discourse-issues-and-perspectives-1st-edition-vijay-k-bhatia/
Historical and Conceptual Issues in Psychology, 3rd

Edition Marc Brysbaert
https://ebookmeta.com/product/historical-and-conceptual-issues-
in-psychology-3rd-edition-marc-brysbaert/
FORMAL LANGUAGE AND AUTOMATA THEORY 2nd Edition Singh

Ajit
https://ebookmeta.com/product/formal-language-and-automata-
theory-2nd-edition-singh-ajit/
Studies in Formal Linguistics Universal Patterns and

Language Specific Parameters Anna Bloch-Rozmej
https://ebookmeta.com/product/studies-in-formal-linguistics-
universal-patterns-and-language-specific-parameters-anna-bloch-
rozmej/
Clarity and Coherence in Academic Writing : Using
Language as a Resource 1st Edition David Nunan
https://ebookmeta.com/product/clarity-and-coherence-in-academic-
writing-using-language-as-a-resource-1st-edition-david-nunan/
Housing Displacement: Conceptual and Methodological

Issues 1st Edition Guy Baeten (Editor)
https://ebookmeta.com/product/housing-displacement-conceptual-
and-methodological-issues-1st-edition-guy-baeten-editor/
Endangered Languages and Languages in Danger Issues of

Documentation Policy and Language Rights 1st Edition
Luna Filipovi■
https://ebookmeta.com/product/endangered-languages-and-languages-
in-danger-issues-of-documentation-policy-and-language-rights-1st-
edition-luna-filipovic/
Power and Truth in Political Discourse Language and

Ideological Narratives 1st Edition Vassil Hristov
Anastassov
https://ebookmeta.com/product/power-and-truth-in-political-
discourse-language-and-ideological-narratives-1st-edition-vassil-
hristov-anastassov/
Handbook of Formal Analysis and Verification in

Cryptography 1st Edition Sedat Akleylek
https://ebookmeta.com/product/handbook-of-formal-analysis-and-
verification-in-cryptography-1st-edition-sedat-akleylek/
Language, Cognition, and Mind
Maxime Amblard
Michel Musiol
Manuel Rebuschi Editors
(In)coherence
of Discourse
Formal and Conceptual Issues
of Language
Volume 10
Editorial Board
Tecumseh Fitch, University of Vienna, Vienna, Austria
Peter Gärdenfors, Lund University, Lund, Sweden
Bart Geurts, Radboud University, Nijmegen, The Netherlands
Noah D. Goodman, Stanford University, Stanford, USA
Robert Ladd, University of Edinburgh, Edinburgh, UK
Dan Lassiter, Stanford University, Stanford, USA
Edouard Machery, Pittsburgh University, Pittsburgh, USA
This series takes the current thinking on topics in linguistics from the theoretical
level to validation through empirical and experimental research. The volumes
published offer insights on research that combines linguistic perspectives from
recently emerging experimental semantics and pragmatics as well as experimental
syntax, phonology, and cross-linguistic psycholinguistics with cognitive science
perspectives on linguistics, psychology, philosophy, artificial intelligence and
neuroscience, and research into the mind, using all the various technical and critical
methods available. The series also publishes cross-linguistic, cross-cultural studies
that focus on finding variations and universals with cognitive validity. The peer
reviewed edited volumes and monographs in this series inform the reader of the
advances made through empirical and experimental research in the language-related
cognitive science disciplines.
For inquiries and submission of proposals authors can contact the Series Editor,
Chungmin Lee at chungminlee55@gmail.com.
More information about this series at http://www.springer.com/series/13376

Maxime Amblard · Michel Musiol ·
Manuel Rebuschi
Editors
(In)coherence of Discourse
Formal and Conceptual Issues of Language
Editors
Maxime Amblard Michel Musiol
LORIA – UMR 7503 ATILF UMR 7118
Université de Lorraine Université de Lorraine
Nancy, France Nancy, France
Manuel Rebuschi
AHP-PReST UMR 7117
Université de Lorraine
Nancy, France
ISSN 2364-4109 ISSN 2364-4117 (electronic)

ISBN 978-3-030-71433-8 ISBN 978-3-030-71434-5 (eBook)
https://doi.org/10.1007/978-3-030-71434-5
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Switzerland AG 2021
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.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Special Reviewers of This Volume
• Kata Balogh, Heinrich Heine University Düsseldorf

• Denis Bonnay, Université Paris Ouest Nanterre
• Francesco Cangemi, Universität zu Köln
• Rachel Cooper, Lancaster University
• Simon Devylder, Lund University
• Richard Dub, Georgia State University
• Bruno Gaume, CNRS
• Charlotte Gauvry, Université de Liège
• Mads Gram, University of Copenhagen
• Julie Hunter, Universitat Pompeu Fabra
• Greg Kobele, Universität Leipzig
• Sylvain Pogodalla, Inria
• Emar Maier, University of Groningen
• Richard Moot, CNRS
• Mehrnoosh Sadrzadeh, Queen Mary, University of London
• Julian Schlöder, University of Amsterdam
• Oriol Valentín, Universitat Politècnica de Catalunya
• Sam Wilkinson, University of Exeter
• Kurt Wischin, Universidad de Granada
v
Invited Speakers and Scientific Committees
of (In)Coherence of Discourse
• Maxime Amblard, Université de Lorraine

• Denis Apotheloz, Université de Lorraine
• Nicholas Asher, CNRS Toulouse
• Valérie Aucouturier, Université Paris Descartes
• Christophe Benzitoun, Université de Lorraine
• Philippe Blache, CNRS, Aix-Marseille University
• Patrick Blackburn, University of Roskilde
• Ellen Breitholtz, University of Gothenburg
• Laurence Danlos, Université Paris 7
• Mathilde Dargnat, Université de Lorraine
• Felicity Deamer, University of Exeter
• Hans van Ditmarsch, CNRS Nancy
• Bart Geurts, University of Nijmegen
• Philippe de Groote, INRIA Nancy
• Klaus von Heusinguer, Universität zu Köln
• Christine Howes, University of Gothenburg
• Jacques Jayez, CNRS Paris
• Hans Kamp, Universität Stuttgart
• Alain Lecomte, Université Paris 8
• Aleksandre Maskharashvili, University of Gothenburg
• Michel Musiol, Université de Lorraine
• Denis Paperno, CNRS Nancy
• Massimo Poesio, Queen Mary University of London
• Sylvain Pogodalla, INRIA Nancy
• Manuel Rebuschi, Université de Lorraine
• Christian Retoré, Université de Montpellier
• Laure Vieu, CNRS Toulouse
• Sam Wilkinson, University of Exeter
vii
Contents
Discourse Coherence—From Psychology to Linguistics and Back

Again . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Maxime Amblard, Michel Musiol, and Manuel Rebuschi
Linguistic and Formal Approaches

Linguistic Recursion and Danish Discourse Particles: Language
in Children with Autism Spectrum Disorder . . . . . . . . . . . . . . . . . . . . . . . . . 21
Patrick Blackburn, Torben Braüner, and Irina Polyanskaya
Reasoning in Multiparty Dialogue Involving Patients
with Schizophrenia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Ellen Breitholtz, Robin Cooper, Christine Howes, and Mary Lavelle
Picturing Questions and Answers—A Formal Approach to SLAM . . . . . 65
Maria Boritchev and Maxime Amblard
Incoherences in Dialogues and their Formalization Focus
on Dialogues with Schizophrenic Individuals . . . . . . . . . . . . . . . . . . . . . . . . . 91
Christophe Fouqueré, Jean-Jacques Pinto, and Myriam Quatrini
Conceptual Issues
Metaphorical Thinking and Delusions in Psychosis . . . . . . . . . . . . . . . . . . . 119
Felicity Deamer and Sam Wilkinson
The Myth of Irrationality: A Wittgensteinian Approach
to Delusions and to the Principle of Charity . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Mathieu Frèrejouan
Incoherent Discourse in Schizophrenia: An Anthropological
Approach to the Mind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Fabrice Louis
Conversations with Madness: Meaning, Context, and Incoherence . . . . . 171
Valérie Aucouturier
ix
Discourse Coherence—From Psychology
to Linguistics and Back Again
Maxime Amblard, Michel Musiol, and Manuel Rebuschi
The present volume explores recent advances in modeling discourse processes,

particularly new approaches aimed at understanding pathological language behavior
specific to schizophrenia. In this work, we examine the modeling paradigm of formal
semantics, which falls within the scope of both linguistics and logic while providing
overlapping links with other fields such as philosophy of language and cognitive
psychology.
This volume is based on results presented during the series of workshops on
(In)Coherence and Discourse organized by Schizophrenia and Language: Analysis
and Modeling (SLAM), a project developed to systemize the study of pathological
language processing by taking an overarching interdisciplinary approach combining
psychology, linguistics, computer science, and philosophy. The principal focus is on
conversations produced by people with psychiatric disorders such as schizophrenia
and autism. The series of (In)Coherence of Discourse workshops hosted by Loria
were held in Nancy, France, in December 2013, December 2014, December 2015,
and March 2017 and include 33 presentations by lead researchers, young researchers,
and invited speakers.
Our focus is on evaluating the heuristics of several formal approaches presented
in the scientific literature aimed at explaining the incongruities and disturbances that
characterize pathological discourse. We examine specifically the degree to which
these formal approaches and experiments support traditional debates about discourse
coherence and related cognitive processes. Based on a comprehensive evaluation of
a range of works that have attempted to explain key disturbances such as disordered
M. Amblard (B)
Loria, Université de Lorraine, CNRS and Inria, Nancy, France
e-mail: maxime.amblard@univ-lorraine.fr
M. Musiol
ATILF, Université de Lorraine, CNRS and Inria, Nancy, France
M. Rebuschi
Archive Henri-Poincaré PReST, Université de Lorraine and CNRS, Nancy, France
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 1

M. Amblard et al. (eds.), (In)coherence of Discourse, Language, Cognition, and Mind 10,
https://doi.org/10.1007/978-3-030-71434-5_1
2 M. Amblard et al.
language and deficit in inferential skills, we show why some approaches are more
relevant than others.
Discourse coherence has traditionally been explored through discipline-specific
research. We propose that considerable potential lies in approaching the question
through a wider lens by considering perspectives across a range of disciplines: notably
linguistics, computer science, philosophy, and psychology. This volume underscores
the importance and value of exploring this particularly complex issue within an
interdisciplinary framework.
In this chapter, we begin by addressing the fundamental differences between
current experimental and clinical approaches. By considering the underlying assump-
tions inherent in each, we show how both approaches fail to account for intentionality
as an independent feature interaction dynamics. In the first section of this chapter,
we show how modeling complex mental operations (including intentional states)
through formal approaches to language can potentially compensate for this lack.
In the second section, we describe several selected theories of linguistics and
dedicate a section to those that have been applied to the study of schizophrenia. In
the third and final section, we revisit the epistemological issues raised by the plurality
of approaches to discourse coherence. We conclude with a more detailed presentation
of the contributions in this volume.
1 Approaches to Psychopathology
Developing systematic analysis methodologies for evaluating verbal communica-

tion skills in patients with schizophrenia, the core focus of the Schizophrenia and
Language: Analysis and Modeling (SLAM) project involves confronting the need
for homogeneity in either corpus or experimental data. The linguistic corpora are
composed of data from neuropsychological tests, selected recordings using eye-
tracking devices or electroencephalography (EEG), and transcriptions of audio
recordings of conversations or clinical interviews involving a person with a psychotic
disorder. These methods of data collection have regularly been used across a
range of scientific studies but are particularly common in the field of cognitive
psychopathology where the combined use of classical experimental methodologies
has given rise to a markedly high number of publications (more than 1,000 arti-
cles per year since 2010, according to PubMed data). Both types of investigations
are more or less guided by the paradigm of cognitive science which, drawing on
aspects of psychology, linguistics, philosophy, and computer modeling, fundamen-
tally seeks to understand the underlying mechanisms of information processing and
cognitive functioning in the brain. Mental illnesses are also traditionally studied
in the context of informal or intuitive clinical settings that take a person-centered
approach and focus on the importance of the client’s subjective worldview. These
approaches were developed largely to respond to the needs of mental healthcare
practitioners and develop more effective outpatient therapy and psychotherapeutic
Discourse Coherence—From Psychology … 3
interventions. The knowledge they produce about the rationality of psychotic disor-
ders is most often reinvested directly back into patient care. Both offer operational
solutions in the immediate sense but do not necessarily meet the requirements of
established scientific methods and are questioned for their validity and even viewed
with suspicion by the scientific community. Any constructive debate or discussion
between the two camps of quantitative and qualitative research representing two very
different world views would be an exercise in futility, even downright impossible,
given the polar differences of their respective underlying theoretical, epistemological,
and methodological assumptions.
In the purely scientific domain, the view of investigating mental illness is a
different matter. Researchers are broadly divided into two schools and favor one of
two approaches that are technically and methodologically complementary. The first is
more classically used in the field of neuroscience and addresses mental illness in terms
of dysfunctional or abnormal biological, psychopharmacological, and neurocogni-
tive processes. It examines the overlap between genetics of the human brain and
what drives thought, or in a broader sense, the processing mechanisms that underlie
thought and behavior (e.g., inference and interpretation, planning and organization
of action). Placed at this end of the spectrum are several investigative techniques
that essentially focus on understanding executive function, memory, attention and,
to some extent, deductive reasoning. This approach is, however, fairly distanced from
the study of thought, language, and communication disorders primarily because it
relies on models more strongly rooted in biology and cognitive psychology, neither
of which are structured to inform language development.
At this level, the traditional framework of neuroscience intersects, at least for
discussion purposes, with a second approach to investigating a mental illness that
falls under the aegis of cognitive psychopathology. It focuses mainly on identifying
neurocognitive disorders (associated with deficits in attention, memory, language,
and executive function, and among others, two commonly measured by neuropsycho-
logical tests being short-term memory and mental flexibility) and understanding the
mechanisms of social cognition, and specifically the processes underlying deficien-
cies in interpersonal relationships. As such, it includes examining cognitive processes
that can be classified as intentional processes while also addressing language deficits
and thought disorders.
Researchers in cognitive psychopathology, however, most often subscribe to a
non-linguistic conception of social cognition. Currently, the most common refer-
ence (and the basis for protocols) is a definition of social cognition proposed roughly
10 years ago at an National Institute of Mental Health (NIMH) Workshop on Social
Cognition in Schizophrenia as “the mental operations that underlie social inter-
actions, including perceiving, interpreting, and generating responses to intentions,
dispositions, and behaviors of others,” (Green et al. 2008). In this context, four core
domains of research are considered to be related to social cognition. In a broad sense,
these are (1) emotional processing or the emotional component focused principally
on the ability to recognize and differentiate basic and complex emotions; (2) social
perception and social knowledge or decoding and interpreting social cues in others;
4 M. Amblard et al.
(3) theory of mind/mental state attribution that encompasses processes of under-

standing, reasoning, and attribution of mental states, regardless if they are emotional
in nature; and (4) attributional style/bias as the way a person explains the causes, or
makes sense, of social events or interactions.
The result is that conventional experimental designs tend to place the question
of language and thought disorders in the background and shift the focus to vari-
ables that are most likely to control linguistic components (phonological, lexical,
or syntactic) that ultimately outside the functioning of linguistic cognition itself,
the very heart of what we call executive functions. Clearly, however, these experi-
mental studies in psychopathology have proven effective, notably in their attempt to
provide new knowledge about the basic operations involved in thought, those that
may be strongly influenced by surrounding neurocognitive and neuropsychological
conditions but are only indirectly related to processes involved in constructing the
semantic representations formulated by the receiver or interlocutor about the mental
state of the speaker.
If we accept the argument put forth by philosophers of language and cognitive
pragmatics that the contextualization processes involved in determining communica-
tive intentions, in discourse or communication, are on one hand based on the ratio-
nality of the cognitive inferential system (the cooperative principle, relevance theory,
and inference rules), and on the other (secondarily, in the chronology of the inter-
pretive mechanism), on specific rules governing contextual access (i.e., pragmatics
and semantic memory), we observe above all else that the experimental method in
cognitive psychopathology has proven more effective in explaining access to context
than in describing the dynamic and inferential constraints of the cognitive interpre-
tive system. Describing the complex intentional processes involved in processing
communication activity requires a methodology that is less static, one which takes
into account the dynamic aspects of the context, which in this case is “discourse”. The
aim is to at once measure the process involved in gaining access to contextual infor-
mation while investigating the dynamics of the inferential process. This experimental
method has failed repeatedly when used to reveal specific cognitive constraints of the
inferential tool, most likely due to the fact that it represents situations being analyzed
in a fixed manner—whereas the distinguishing feature of the cognitive tool is that it
is dynamic.
Conversely, despite the holistic nature of the contextual information rendered
accessible through the inferential mechanism, formal methods long considered as
reliable in linguistic analysis have proved more effective in capturing the rationality
of the inferential process itself. These methods have been shown to be more successful
in not only representing the dynamic character of the communication action but in
modeling variables and properties involved in an interaction that contribute to action
planning. Ultimately, this makes it possible to isolate and identify selected features
and compare the degree to which they activate or inhibit the inferential mechanism.
Moreover, since the famous “dynamic turn” in the 1990s, linguistic analysis and
modeling in formal semantics have been based on dynamic discourse movements
and not solely on fixed questions, which represents a remarkable step forward in the
analysis of intentional processes. In this sense, this type of methodology effectively
compensates for defects associates with the experimental method and facilitates
investigating the dynamic properties of the inferential mechanism beyond the overly
simplified explanation of reactive and automated processes.
2 Approaches to Language and the Linguistic Approach

to Schizophrenia
Cognitive Psychotherapy treats verbal material more abstractly and independently

of linguistic and discursive factors by using models and theories that do not account
for context or the situational conditions in which the verbal material is collected (i.e.,
clinical interviews or conversations). It tends to reduce explanations of communica-
tion difficulties exhibited by patients with schizophrenia to executive dysfunction,
and by association neuropsychological dysfunction. Linguistic elements involved,
or language disorders, are at best assessed outside the context of discourse (and
therefore, outside of the communication context) without reference to language
theory. Added, the processes for analyzing communicative intentions, for example,
are evaluated in a wider sense under social cognition without taking into account
the specific inferential mechanism (inference rules) involved in recognizing commu-
nicative intentions (those that are implicit, for example). Difficulties in schizophrenia
patients are attributed directly to simple executive functions or, more broadly and
even less precisely, to problems of semantic memory.
Nevertheless, the implicit conceptions of the relationship between language and
thought that the experimental protocols used so widely in linguistic research are
based on remain close to the original concept of Bleuler (1911) that language is
a direct expression of thought. Disturbances of thought or thought disorders have,
thus, been a subject of high interest in psychiatry and psychoanalysis from the early
1900s. It was not until the 1970s that the field of linguistics would contribute to
the question through the development of more formalized analytical methodologies
(Chaika 1974; Fromkin 1975). The debate shifted course at this point to the question
of whether or not disturbances or language abnormalities exhibited by schizophrenia
patients in interviews or in their general communication are more closely connected
to deficits in language competence or to thought disorder.
These initial debates laid an important foundation for what we currently know
about the diagnostic process, our understanding of specific disorders and advances
in psychotherapeutic patient care, and yet they are today, around the world, only
modestly addressed in professional training in psychology, and even less so in
psychiatry. Additionally, psychologists and psychiatrists rarely receive training in
language science which means that professionals and practitioners in both fields are
equipped with a limited understanding of the importance of systematic data collec-
tion and corpus development. Data collection largely continues to take place on-site
in clinics or settings where patients receive care. While defining protocol guide-
lines for compiling corpus data is fairly straightforward, the quality of results can be
6 M. Amblard et al.
highly variable; this is due in particular to the specificity of the subject under study,
the nature of mental disorders that inherently render social interactions delicate and
unstable, and to patients’ difficulties adhering to restrictive or structured approaches
these assessment strategies require.
As we previously mentioned, mental disorders are frequently studied through their
manifestations as language abnormalities. Deepening our understanding of the mech-
anisms underlying disordered discourse should, naturally, be approached through
linguistics. An obvious point of departure is studying and analyzing the singular
aspect of language production during interactions with patients with schizophrenia
(Kuperberg 2010a, b) and conducting a comprehensive meta-analysis of the liter-
ature on linguistic approaches to these abnormalities in discourse production that
highlights three types of characterization viewed through the lens of linguistics.
The first focuses on a frequency analysis of data on language and measuring
predictability by using Cloze Analysis (Taylor 1953) and variability of a specific
word in its context with Type Token Ratio (TTR). This type of analysis favors,
for example, the notion that schizophrenia patients have reduced competence in
mastering linguistic complexity (e.g., experience greater difficulty in predicting
linguistic content) (Salzinger et al. 1970, 1979). Previous automatic analyses have
shown that schizophrenics produce more associations between words (Maher et al.
2005), which would support the argument that shifts in themes occur with greater
frequency. In addition, a study by Elvevåg et al. (2007) examined the lexical dives
observed in transcribed interviews with a patient suffering from schizophrenia and
showed that they produced much lower scores in a semantic association, supporting
the argument that they produce less-coherent discourse.
The second approach focuses on syntactic and lexical structures, along the same
lines taken by Chaika (1974) and Fromkin (1975). A number of previous studies
have successfully shown that schizophrenia patients are less capable of dealing with
syntactic complexity (Fraser et al. 1986) and tend to use incorrect grammar (Hoffman
and Sledge 1988), most commonly seen in their simplified sentence construction.
This would suggest that these patients would also have difficulty understanding
sentences with complex syntactic construction. What has emerged fairly clearly
is that this lower syntactic complexity, both in terms of production and compre-
hension, is a telling characteristic of schizophrenia, while these same patients are
able to produce and understand simple phrasal structures (Thomas et al. 1990).
Importantly, patients with schizophrenia rarely display signs of syntactic impair-
ment. Even the most incoherent statements or the “salad of words” are in fact
composed of normal syntactic elements (Andreasen 1979), suggesting that even
markedly reduced language performance is not necessarily indicative of a concrete
syntactic deficit. According to Covington et al. (2005), it is more likely that abnormal-
ities in schizophrenic speech point to memory and information-processing deficits
or cognitive disturbance (i.e., attention deficit) as constraints impacting communica-
tion rather than to a primary defect in linguistic competence (i.e., disrupted syntactic
processing).
Finally, the third approach focuses on the linguistic interpretation of discourse. The
objective here is to emphasize the skills involved in constructing normal interactions
rather than analyzing the referential processes themselves (linking anaphors to their
antecedents) (originally Rochester and Martin 1979; Ditman and Kuperberg 2010).
Docherty et al. (1996) developed a comprehensive measure to capture a range of refer-
ential communication failures including vague, confused, and missing references.
This work essentially examines anomalies in speech production from a linguistics
perspective by qualitatively measuring the semantic relationships involved. As for
the dynamics of controlling enunciation, or the distinctiveness of utterances, several
studies have reported observing in some patients with schizophrenia impaired affec-
tive prosody; this refers to the neuropsychological function encompassing all non-
verbal aspects of language necessary for both recognizing and conveying emotions
in communication. According to Cutting (1979), the percentage of patients with
schizophrenia who exhibit a deficit in either perceiving or expressing emotional
prosody may be as high as 85%. It has also been shown that acoustic measures of vocal
expression and production are attenuated in patients suffering from schizophrenia
who also experience affective anesthesia, or a blocking of affective responses (one of
the more negative symptoms associated with the condition) yet increase in patients
with schizophrenia who do not experience psychological blocking off. The ampli-
tude (volume) and frequency (pitch) of utterances by individuals suffering from both
schizophrenia and affective disorders have been shown to be lower than they are in
both normal participants and individuals with schizophrenia without affective anes-
thesia (Alpert et al. 1989). Therefore, these authors posit that we would expect the
production of emotional prosody to be disrupted in patients with schizophrenia, while
the level of linguistic content may be preserved.
In general terms, as Covington et al. (2005) and others have argued, the most
pronounced disruption in schizophrenia occurs at the pragmatic level, essentially the
point where language production and context processing meet. The most outward sign
of dysfunction in schizophrenia often manifests in disordered speech or unexpected
or unintelligible outbursts; an individual living with schizophrenia will say strange
things that are contextually inappropriate. Other authors share this view, notably
since Frith (1988) established that capacities in schizophrenia patients’ syntactic and
semantic processing can be unaffected while the same patients will have difficulty
with more complex use of language, or what is known as pragmatics.
Interestingly, all of these approaches consider verbal expression in schizophrenia
as an important vector for understanding thought disorder. This relatively strong
assumption appears to be widely accepted and could considerably widen the scope
of exploration, except that thought disorder itself has received only limited attention
in the literature; thought disorder, or formal thought disorder, has historically been
described somewhat broadly as the complex link between thought, language, and
speech. One obvious problem is that addressing language output in schizophrenia
necessarily requires considering the specificities of the patient concerned to gauge
the level of disorder in their language and assign those abnormalities to levels defined
by a Saussurian vision of language. A second difficulty associated with the line of
research, also pointed out by Kuperberg, is that patients suffering from schizophrenia
are either under the influence of prescription antipsychotic medication, or they have
stopped taking their medication and their symptoms consequently become more
8 M. Amblard et al.
severe, which makes interaction even more difficult if not outright impossible. It is
therefore extremely difficult to determine if language abnormalities are symptoms of
the disease, or if they are related to the effects of the patient’s medication. This may
partially explain the fact that a number of previous works have reported syntactic
impairment in schizophrenia, which our own more recent morphosyntactic analysis
did not support (Amblard et al. 2015). What is undeniably clear is that schizophrenic
discourse is structured very differently than normal discourse and necessarily requires
approaches to address this singular difference. It is, however, important to note that
unlike language abnormalities considered symptomatic of other disorders, such as
asphasia, anomalies in schizophrenic speech selectively involve levels of linguistic
structure which can vary from the lower semantic level to the more complex pragmatic
level characterized by higher degrees of abstraction in terms of conceptual language
modeling, which effectively are very difficult to identify.
This dynamic has been as much a subject of interest in linguistics as it has
in computational linguistics. These two fields combined have produced a signifi-
cant number of recent advances in both language production theories and language
processing tools for studying speech patterns in schizophrenia, but they approach the
question from two distinctive angles.
The first looks to computational linguistics for developing language processing
tools and computerized models adapted to analyze speech production data collected
from schizophrenia patients. Researchers at IBM (Bedi et al. 2015) achieved this
by combining automated speech analysis and a machine learning algorithm as a
classifier to evaluate a corpus of transcripts for semantic and syntactic complexity.
The results of this study are impressive in that they show how computerized methods
can be used to characterize complex behavior such as speech and ultimately can
identify and predict schizophrenia. The downside is that this study only included
5 patients with schizophrenia, which underlines an important point—as much as
these types of approaches open novel avenues of study, the reasoning of moving
toward strictly implementing them as protocols has been questioned simply because
sufficient quantities of relevant data are highly difficult to obtain.
The other approach or interest in combining linguistics and computational linguis-
tics places focus on the models themselves. Ultimately, the aim is to produce soundly
defined formal representations that are both cognitively motivated and justified in
the linguistic framework. This calls to mind Montague semantics (Montague 1974),
a theory of natural language semantics and of the relationship with syntax, as well
as those approaches that integrate the dynamic aspects of discourse interpretation
(Kamp and Ryle 1993) or extensions of Segmented DRT (Asher and Lascarides
2003). These models, among others, are generally the basis for implementing natural
language processing. Thus, studying how accurately they reflect atypical speech
patterns can by extension provide valuable information about what can or cannot be
automatically identified. This implies the need to consider a normative context when
determining or envisaging possible implications of these models.
Indeed, a legitimate line of questioning is whether, or to what degree, these models
can accurately represent a language in the absence of constraints. This language
would necessarily become the place where cognitive processes take place. There-
fore, models that are capable in principle of differentiating between standard and
unconventional uses of language would ideally simulate these processes. This in turn
would validate the reality represented by the language recognized by these models
and would support the cognitive validity of the models themselves. This should still
be true without specifying or defining processes involved in language acquisition.
Cognitive reality in the context of computational modeling, however, is difficult to
define. Considering these incongruities allows us to compare these models through
complex phenomena present in the language providing more concrete validation.
As we alluded to earlier, incongruities that emerge when communicating with
schizophrenia patients occur at a highly abstract level situated between semantics
and pragmatics. The fact that only limited empirical evidence has been reported
in the literature on deficits in abstraction, in our view, makes it critically impor-
tant to direct more attention to this defining aspect of language dysfunction in
schizophrenia, and underlines the importance of integrating “abstraction” as a crite-
rion in formal language models. Lastly, given the highly specific nature of language
in schizophrenia, more emphasis should be given to developing accurate representa-
tions of dialogue analyzed by processing models, than evaluating the validity of the
models themselves.
3 Contributions of Pluralism
Coherence and incoherence are normative concepts. Identifying inconsistencies in

discourse, however, does not necessarily call for a high-level metalinguistic assess-
ment. Most speakers are generally able to grasp when things are not going well or
when a conversation begins to veer off course. They can sense a change in register, or
style of language, to signal humor, poetry, or “madness”. This sensitivity is acquired
with linguistic competence: ensuring that we are understood in verbal interactions
involves not only accommodating contributions made by others, but also integrating
intentions and content communicated as common ground or shared reference and
making the necessary adjustments when the speaker makes a mistake.
The philosophers Quine (1960) and Davidson (1970, 1980) insisted that under-
standing, interpreting, and responding to the arguments of others relies heavily on the
principle of charity: interpretation of the beliefs and utterances of others that maxi-
mizes the truth or rationality of what others think and say, without which communica-
tion is impossible. Underlying this strategy, to use Wittgenstein’s own terminology, is
essentially a “language game” rooted in shared conceptual parameters that takes place
in a cultural environment both parties understand, a common “form of life”. This
fundamentally describes how sharing the same form of life and the same language
game can lead us to implicitly assume that we share the same rational norms with our
interlocutor and, effortlessly, we are able to understand the meaning of our linguistic
exchanges.
10 M. Amblard et al.
The immediacy of understanding can be at risk when others violate the rules
expected of us in communication. In the absence of a shared “common ground” of
syntax when communicating with a person who does not master the language being
used, or when the lexicon and semantics diverge, understanding is rendered more
difficult or even impossible. The presence of madness in verbal exchange is one of the
more prominent sources of conflict. Some consider, along the same lines as Jaspers
(1913), that psychosis is characterized by that which is incomprehensible, which
effectively distinguishes it from simple neurosis. Understanding is fundamentally
impossible in psychosis due to the lack or failure of empathy (Einfühlung): I cannot
see myself in this person suffering from psychosis, so therefore I cannot understand
him or her.
Importantly, the fact that we stop understanding does not prevent us from trying
to explain, which leads us back to the classic distinction drawn by Dilthey (1989)
between the science of the mind and the natural sciences. While Freud notori-
ously contested this position by arguing that interpretation could effectively replace
empathy when attempting to understand a person exhibiting psychosis, the more
widely held approach to madness reduces its apprehension to naturalism. However,
whether it comes down to a question of favoring neurobiological, neurochemical or
even genetic causes, or all of the above combined with multifactorial explanations,
the naturalist paradigm is limited to a third-person approach in that it eliminates the
subject’s perspective altogether. The reasons, motives, and rationality of the patient
suffering from psychosis are simply not considered relevant. This position, however,
is problematic not only for the person suffering from psychosis essentially withdrawn
from the equation, but also for the scope of the analysis.
Indeed, incoherent and disordered speech characterized by frequent derailments
are, at the linguistic level, among the most frequently observed symptoms of
schizophrenia. Assessments of disordered speech or contradictory statements neces-
sarily require relying on rational norms rather than natural laws. Similarly, the occur-
rence of delusions, hallucinations, or general denial of reality commonly associated
with schizophrenia can only effectively be detected by evaluating semantic coherence
of thought contents. The naturalistic approach offers no way of analyzing the rational
and semantic dimensions of schizophrenic thinking. Understanding the rationality of
a person with schizophrenia is possible, however, through linguistic interaction by
means of normative modeling, or in other words by using theoretical tools developed
by the fields of logic and formal linguistics.
The plurality of theoretical approaches to studying disordered discourse thus
requires going beyond the mere integration of naturalistic explanations. It becomes
a question not only of combining different structural levels of mental illness and
cognitive processes, but also of accepting models that cannot be definitively inte-
grated because they involve other levels. A pluralism that stops at recognizing the
diverse nature of biological, neurochemical, and neuropsychological factors would
not be sufficient. In psycholinguistics, we must instead assume a conceptual pluralism
based on the essential difference between factors that can be described in natural-
istic terms, and linguistic and rational factors that can be analyzed by normative
approaches.
Studies on discourse incoherence commonly use corpora involving patients diag-

nosed with mental disorders and generally rely on a distribution of interlocutors
described as either patients or normal controls, a division that is simply not factual.
This distribution presupposes a positioning of the boundary between normal and
pathological which cannot be defined in biological terms, as evidenced by debates
surrounding the classification system proposed by successive editions of the DSM.
The criteria that separate what is considered normal from pathological in terms of
mental suffering are socially constructed, and subsequently, this means that they are
far from arbitrary categories.
The scientific identification of discourse inconsistencies thus implies a crossover
of several normative systems: those that define the boundaries of the normal and
the pathological (neurobiological or psycho-cognitive) and those that constitute the
grammar of coherent verbal communication (lexicon, syntax, semantics, and prag-
matics). Pluralism ultimately appears to represent the reverse of another aspect of
scientific knowledge, first theorized by (Duhem 1906) and later by (Quine 1951),
being holism. Faced with the inherent problem of inconsistencies in discourse, virtu-
ally no discipline works with isolated data, whereas theoretical models regularly do
due to an essential mutual interdependence in normative delimitations. This holism
suggests that striving for greater balance between disciplines in terms of scientific
procedure would allow each to rely on the results of others as hypotheses to be tested.
Through its theme and contributions, the present work should therefore be read
as an illustration of how conceptual holism and pluralism play an essential role in
relationships developed between the fields of logic, linguistics, psycholinguistics,
and cognitive psychology as each of these disciplines confront the challenge of
understanding a unique form of the abnormal.
4 Presentation of the Volume
The volume is divided into two parts. The first part presents works that explore
linguistics and formal semantics and pragmatics, all of which deal with patholog-
ical conversational contexts, either in the presence of autism (the first chapter) or
schizophrenia (the following chapters). The second part presents a collection of less
formal and more conceptual studies that examine different approaches to under-
standing madness, delirium, and irrationality, prompting several authors to turn to
Wittgenstein’s thinking as to the nature of language.
Part 1. Linguistic and Formal Approaches.
The first part of this volume draws on formal models of language production in an
interdisciplinary context. It focuses on a twofold movement. On the one hand, it
highlights the capacity of these models to identify interactions that do not follow a
standard form and provide a foundation for developing arguments to explain this devi-
ation from the norm. On the other hand, these studies give substance to these models
in the form of empirical and cognitive validation supporting the capacity to model
natural phenomena. The first chapter looks at the case of autism and projections of
“false beliefs”. The rest of this section focuses on the modeling conversations that
involve patients with schizophrenia. The second article analyses language produc-
tion in schizophrenia in a relatively free discussion task based on a very abstract
semantic level as the topoi. The last two articles focus on language production in
conversational exchanges between patients and psychologists. One defines semantic
representations based on higher order logic, the other models them based on play-
fulness, a specific logic for evidence and evidence against that which gives meaning
to dialogue. Taken together, these studies emphasize the potential in taking a formal
approach to interpreting interaction.
The first chapter is entitled “Linguistic Recursion and Danish Discourse Particles:
Language in Children with Autism Spectrum Disorder”, written by Patrick Black-
burn, Torben Braüner, and Irina Raheemah. In this chapter, the authors describe their
research on second-order false beliefs (SOFB) in children with autism syndrome.
This interdisciplinary work, in the truest sense, associates experimental method-
ology taken from psycholinguistics and psychology with linguistic models, which
these authors developed further by integrating concepts borrowed from artificial
intelligence and epistemic logic. The scientific literature, they point out, has to date
largely focused on first-order false beliefs, which this work attempts to address: they
offer an updated perspective and discussion about the relationship between the level
of control of SOFB and the control of language functioning in autistic children. By
using modal particles in Danish based on the JDV test and the Recursive Embed-
ding Tool, they show that good control of recursive embedding increases a person’s
ability to recognize false beliefs. They analyze the reasoning processes underlying
the two tests and highlight the common elements related to reasoning identified at
the level of particle control in Danish (JDV) as this facilitates recognizing false
beliefs. They extend their analysis and reflections to consider neurotypical children
and their stages of their ontogenetic development, and more generally, the question
of acquiring beliefs.
In their chapter entitled “Reasoning in Multiparty Dialogue Involving Patients
with Schizophrenia”, Ellen Breitholtz, Robin Cooper, Christine Howes, and Mary
Lavelle present their work on the formalization of reasoning, and focus specifically on
developing formal models of arguments that appear in conversations involving indi-
viduals with schizophrenia. Particular attention is given to their report of an excerpt
from a conversation sourced from the DRiPS corpus, a body of transcripts of discus-
sions between three people (including one person diagnosed with schizophrenia)
who discuss the moral dilemma of the balloon: passengers in a hot air balloon realize
that one person will need to leave if any of them are to survive, but how do they decide
who? This situation requires the participants to share not only information, but also
their own reasoning processes. The authors rely principally on the concepts of topoi
and enthymemes as representing general principles of reasoning and their instantia-
tion through verbal exchanges. These interactions are managed through a dialogue
gameboard within the framework of the Type Theory with Records and the prin-
ciples of game theory. This methodological tool, these authors posit, helps explain
differences in reasoning supported by non-logical inferences leading to misunder-

standings that require new accommodations to become understandable.
The research presented by Maria Borichev and Maxime Amblard in the chapter
“Picturing Questions and Answers—A Formal Approach to SLAM” concerns the
compositional treatment of assertions, questions, and responses in dialogue. These
authors describe using formal methods in their analysis configured to model dialogue
in a dynamic framework, with the specific focus being assessing recursive and
sequential properties of conversation, in normal or pathological contexts. In their
analysis of empirical data from the SLAM corpus (mentioned above), they base
formal constructions largely on frame semantics, which allowed for a contextual
update of linguistic expressions specific to negotiation in dialogic interactions. They
were able to develop a frame semantic taxonomy of questions and a frame semantic
description of negation, and propose a description of dynamic operators involved in
calculating the information being negotiated and integrated at the level of conver-
sational patterns pairing, for example, “assertion–assertion”, “assertion–question”
and “question–answer”. In addition, they explain how the different formal models
presented in their work show high potential for adequately explaining abnormal-
ities in schizophrenia at the discourse level, and finally, describe conditions that
would facilitate developing a future formal model with the capacity to recognize
conversation ruptures in normal conversation.
The last chapter of Part One features a work entitled “Incoherences in Dialogues
and their Formalization Focus on Dialogues with Schizophrenic Individuals” by
Christophe Fouqueré, Jean-Jacques Pinto, and Myriam Quatrini who describe their
use of a relatively recent extension of logic. First developed by Jean-Yves Girard
around 2000, Ludics is a logical framework wherein interaction is a central concept
and takes place between two designs as a step-by-step travel through two dual paths,
with one path in each design being a set of potential paths where interaction may
take place. The same way the concepts of logic relate naturally to philosophy and
language, this novel approach lends itself to the nature of dialogue and linguistic
representations. In Ludics applied to the dynamics of dialogue, playfulness as a
level of linguistic expressiveness produces two parallel representations, which in the
context of conversation correspond to the person speaking and the person listening
(the receiver). In this work, the authors show how the two viewpoints of the conver-
sation can veer in different directions yet remain rational within the model. It is these
divergences that lead to misunderstandings, which they seldom do as this would
violate the rules of logic. Logical coherence, therefore, can be understood in terms
of cohesion. These authors illustrate these ruptures of cohesion through excerpts
taken from conversations involving patients with schizophrenia.
Part 2. Conceptual Issues.
The second part of this volume looks at our general understanding of what happens
when a patient with schizophrenia disrupts a conversational interaction. The first
chapter focuses on the articulation between metaphorical and literal meanings, which
requires certain mastery in the conversational context. This is where people with
schizophrenia appear to have the greatest trouble. The question of insanity or irra-
tionality thus extends beyond just “the head”, but manifests in the course of general
interaction, in conversational exchanges in particular. In this respect, Wittgenstein’s
philosophy proves to be a valuable contribution. Breaking with the internalism of
the mind, Wittgenstein offers a perspective in which expression only becomes mean-
ingful in the context of language games and a shared form of life. From this perspec-
tive, Wittgenstein invites us to approach understanding coherence in discourse by
considering the context in a broader sense, which the next three chapters illustrate.
Felicity Deamer and Sam Wilkinson focus on the causal mechanism of delirium
in psychosis. Their work “Metaphorical Thinking and Delusions in Psychosis” exa-
mines the hypothesis that the experience of inner speech resonates with the rationality
of the metaphorical thought process, which contributes at once to the emergence of
delirium and to the development of its content. Thus, it is the literal meaning of
a metaphorical expression that can be extracted from the proposal with metaphor-
ical meaning, and which would, despite its context of enunciation, relates directly
to the occurrence of delusional, bizarre, or colorful assertions (Finbarr delusions).
The authors ascribe this singular mechanism for understanding spoken sentences to
cognitive failure, commonly found in delusional schizophrenic patients, which in
this context often presents as a difficulty in inhibiting the literal meaning despite the
context of enunciation. Interestingly, they also explore the experience of delirium
from the patient’s perspective and advance the hypothesis that while patients with
schizophrenia may experience inner speech as assertive and sincere, the content of
their thoughts ultimately remains separate from reality and has no real consequences.
In his article “The Myth of Irrationality: A Wittgensteinian Approach to Delusions
and to the Principle of Charity”, Mathieu Frerejouan discusses the merits and limits
of relying on the principle of charity to recognize the irrational. Can this principle be
applied to delusional speech, and if not, how can we understand the latter? The philos-
ophy of psychiatry tends to oscillate between the systematic postulate that applying
the principle of charity influences the rationality of the patient with delusion, and
the symmetrical rejection that rationality can be recast by this principle. The author
proposes that understanding rationality needs to be considered from a wider angle
than one based purely on logic, notably by incorporating the rules governing a course
of action that essentially represents Wittgenstein’s basic or “hinge” certainties. And
as such, rather than interpreting specific delusions or irrationality itself by attempting
to understand the illusory nature of each, we might consider how each aligns to our
norms of what is rational and, therefore, meaningless.
Fabrice Louis’ article “Incoherent Discourse Schizophrenia: An Anthropological
Approach to the Mind” also alludes to Wittgenstein but turns instead to his thinking on
the concepts of the language game and life form. Here, the author underlines the value
of adjusting our beliefs and taking a more pragmatist, which ultimately led him to
attribute incoherent utterances in schizophrenic speech as manifestations of problems
related more closely to interaction than to semantics. Unlike the previous work that
considers delusional discourse as meaningless, Louis suggests that it can be assigned
meaning interpreted in the context of another form of a language game specific to the
individual with schizophrenia. Utterances in a more ordinary language game that may
count as an assertion or expression of beliefs may, in another, signify an expression

of emotion. This shift in the nature of the game between interlocutors produces
conversational disturbance, and ultimately isolates the schizophrenic speaker, even
without considering semantic inconsistencies.
Part Two closes with a chapter by Valérie Aucouturier, “Conversations with
Madness: Meaning, Context and Incoherence”. This author focuses here on
Wittgenstein’s concept of shared “form of life”—being the common humanity in
language—between normal speakers and individuals with schizophrenia. Conversa-
tional ruptures that arise against this common backdrop are not, the author argues,
necessarily due to a breakdown of rationality on the part of the schizophrenic speaker
but instead stem from a gap between the two interlocutors’ reference contexts. The
construct of the schizophrenic speaker’s perspective relies less on the principle of
charity as a projection of rationality perceived in the other’s speech but is more
reflective of their ability to understand the interlocutor’s context references, their
beliefs, and as a result, the meaning being conveyed. This chapter thus concludes
on an optimistic note being that inter-comprehension in conversation in the presence
of madness is achievable if we look beyond the linguistic, semantic, and pragmatic
aspects toward focusing on understanding and modeling the broader context of the
exchange.
Acknowledgements This work was supported partly by the French PIA project “Lorraine Univer-
sité d’Excellence”, reference ANR-15-IDEX-04-LUE. We thank all the reviewers of the volume
and of each chapter. We also thank Aimee Orsini for her proof-reading which is, as always, of
great quality. We finally thank the members of the scientific committee and the reviewers of the
(In)coherence of Discourse workshops for their valuable support.
References
Alpert, M., Rosen, A., Welkowitz, J., Sobin, C. (1989). Vocal acoustic correlates of flat affect in
schizophrenia: Similarity to Parkinson’s disease and right hemisphere disease and contract with
depression. The British Journal of Psychiatry, 154(4), 51–56.
Amblard, M., Fort, K., Demily, C., Franck, N., & Musiol, M. (2015). Analyse lexicale outillée de la
parole transcrite de patients schizophrènes. Traitement Automatique des Langues, 55(3), 91–115.
Andreasen, N. C. (1979). Thought, language and communication disorders. I. Clinical assessment,
definition of terms, and evaluation of their reliability. Archives of General Psychiatry, 36, 1315–21.
Asher, N., & Lascarides, A. (2003). Logics of conversation, studies in natural language processing.
Cambridge University Press.
Bedi, G., Carrillo, F., Cecchi, G. A., Slezak, D. F., Sigman, M., Mota, N. B., Ribeiro, S., Javitt, D.
C., Copelli, M., & Corcoran, C. M. (2015). Automated analysis of free speech predicts psychosis
onset in high-risk youths. Npj Schizophrenia, 1, 15030 EP— 08.
Bleuler, E. (1911). Dementia praecox, or the group of schizophrenias. New York: International
Universities Press.
Chaika, E. (1974). A linguist looks at ‘schizophrenic’ language. Brain and Language, 1, 257–276.
Covington, M., He, C. Brown, C. Naci, L., McClain, J., Fjordbak, B., Semple, J., & Brown, J. (2005).
Schizophrenia and the structure of language: The linguist’s view. Schizophrenia Research, 77(1),
85–98.
Cutting, J. (1979). Memory in functional psychosis. Journal of Neurology, Neurosurgery and

Psychiatry, 42(11), 1031–1037.
Davidson, D. (1970). Mental events. In L. Foster & J. W. Swanson (Eds.), Experience and theory.
London: Duckworth. Repr. in Davidson, D. (1980). Essays on actions and events. Oxford: Oxford
University Press.
Dilthey, W. (1989). Einleitung in die Geisteswissenschaften (Introduction to the Human Sciences)
(Volume I). Repr. in Selected works, Makkreel, R. A., & Rodi, F. (Eds.). Princeton, NJ: Princeton
University Press (vol.I, 47–242, 1883).
Ditman, T., Kuperberg, G. R. (2010). Building coherence: A framework for exploring the breakdown
of links across clause boundaries in schizophrenia. Journal of Neurolinguistics “Language in
Schizophrenia,” (Special Issue).
Docherty, N. M., DeRos, M., & Andreasen, N. C. (1996). Communication disturbances in
schizophrenia and mania. Archives of General Psychiatry, 53, 358–364.
Duhem, P. (1906). La Théorie physique, son objet, sa structure. Paris: Chevalier et Rivière.
Elvevåg, B., Foltz, P. W., Weinberger, D. R., Goldberg, T. E. (2007). Quantifying incoherence
in speech: An automated methodology and novel application to schizophrenia. Schizophrenia
Research, 93(1–3), 304–316. https://doi.org/10.1016/j.schres.2007.03.001.
Fraser, W. I., King, K., & Thomas, P. (1986). The diagnosis of schizophrenia by language analysis.
The British Journal of Psychiatry, 148, 275–278.
Frith, I. C. D. (1988). Cognitive Neuropsychology of schizophrenia. Howe: LEA, 1992.
Fromkin, V. A. (1975). A linguist looks at ‘schizophrenic’ language. Brain and Language, 2,
498–503.
Green, M. F., Penn, D. L., Bentall, R., Carpenter, W. T., Gaebel, W., Gur, R. C., & Heinssen,
R. (2008). Social cognition in schizophrenia: an NIMH workshop on definitions, assessment,
and research opportunities. Schizophrenia Bulletin, 34(6), 1211–1220. https://doi.org/10.1093/
schbul/sbm145.
Hoffman, R. E., & Sledge, W. (1988). An analysis of grammatical deviance occurring in spontaneous
schizophrenic speech. Journal of Neurolinguistics, 3, 89–101.
Jaspers, K. (1913). Allgemeine Psychopathologie, Ein Leitfaden für Studierende, Ärzte und
Psychologen (1st ed.). Berlin: Springer.
Kamp, H., & Reyle, U. (1993). From discourse to logic. Dordrecht: Kluwer Academic.
Kuperberg, G. R. (2010a, August). Language in schizophrenia part 1: An introduction. Lang Linguist
Compass, 4(8), 576–589. https://doi.org/10.1111/j.1749-818X.2010.00216.x,2010.
Kuperberg, G. R. (2010b). Language in schizophrenia part 2: What can psycholinguistics bring to
the study of schizophrenia... and vice versa ? Lang Linguist Compass, 4(8), 590–604. https://doi.
org/10.1111/j.1749-818X.2010.00217.x.
Maher, B. A., Manschrec, T. C., Linne, J., & Candela, S. (2005). Quantitative assessment of the
frequency of normal associations in the utterances of schizophrenia patients and healthy controls.
Schizophrenia Research (78), 219–224.
Montague, R. (1974). Formal philosophy: Selected papers of Montague. Yale University Press.
Quine, W. V. O. (1951). Two dogmas of empiricism. The Philosophical Review, 60(1), 20–43.
https://doi.org/10.2307/2181906.
Quine, W. V. O. (1960). Word and object. Cambridge, MA: MIT Press.
Rochester, S., & Martin, J. R. (1979). Crazy talk: A study of the discourse of Schizophrenic speakers.
New York: Plenum Press.
Salzinger, K., Pisoni, D. B., Portnoy, S., & Feldman, R. S. (1970). The immediacy hypothesis
and response-produced stimuli in schizophrenic speech. Journal of Abnormal Psychology, 76,
258–264.
Salzinger, K., Portnoy, S., & Feldman, R. S. (1979). The predictability of speech in schizophrenic
patients [letter]. The British Journal of Psychiatry, 135, 284–287.
Taylor, W. (1953). ‘Cloze’ procedure: A new tool for measuring readability. Journalism Quarterly,
30, 415–433.
Thomas, P., King, K., Fraser, W. I., & Kendell, R. E. (1990). Linguistic performance in
schizophrenia: A comparison of acute and chronic patients. The British Journal of Psychiatry,
156, 204–210.
Linguistic and Formal Approaches
Linguistic Recursion and Danish
Discourse Particles: Language
in Children with Autism Spectrum
Disorder
Patrick Blackburn, Torben Braüner, and Irina Polyanskaya
Abstract In a study involving 62 Danish children with autism spectrum disorder,

we obtained results showing that the mastery of linguistic recursion is a significant
predictor of success in second-order false belief tasks. The same study also showed
that the mastery of linguistic recursion was not significantly correlated with success
in a task involving three heavily used Danish discourse particles. This calls for fur-
ther explanation, as the reasoning involved in both types of tasks seems similar. In
this paper, we discuss second-order false belief reasoning, the reasoning underly-
ing the use of the three Danish discourse particles, say what we know about them
experimentally, and discuss what they do (and don’t) have in common.
Keywords Recursion · Autism spectrum disorder · Sentential complements ·

Discourse particles · Theory of mind · Second-order false beliefs · Logic
1 Introduction
In the first part of this paper, we describe the theoretical background, experimental
design, and a key result of a recently completed study (Polyanskaya et al. 2018;
Polyanskaya 2018) involving 62 Danish children formally diagnosed with Autism
Spectrum Disorder (ASD): the mastery of linguistic recursion (and in particular,
sentential complementation) is a significant predictor of success in second-order false
belief tasks for children with ASD. We view this result as an endorsement of an idea
proposed in Tager-Flusberg and Joseph (2005), namely that children with ASD use
language in general, and sentential complements in particular, to “bootstrap” their
P. Blackburn (B)
Centre for Communication and Arts, Roskilde University, Roskilde, Denmark
e-mail: patrick.rowan.blackburn@gmail.com
T. Braüner · I. Polyanskaya
Department of People and Technology, Roskilde University, Roskilde, Denmark
e-mail: torben@ruc.dk
I. Polyanskaya
e-mail: irinapol.translations@gmail.com
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 21
M. Amblard et al. (eds.), (In)coherence of Discourse, Language, Cognition, and Mind 10,
https://doi.org/10.1007/978-3-030-71434-5_2
22 P. Blackburn et al.
false belief understanding. Or to adopt the terminology used in Lind and Bowler
(2009), children with ASD may use language to “hack out” solutions to false belief
tasks.
In the second half of the paper, we contrast this result with some others from the
same study. These other results concern the JDV test, a gap-filling task, introduced
in Thomsen (2012), Engberg-Pedersen and Thomsen (2012), involving three heavily
used Danish discourse particles: jo, da, and vel. The JDV test was designed to be a
language test requiring social cognition skills. It was used by its designers in a study
of Danish-speaking children with ASD; the study showed that children with ASD
score lower on the JDV test than typically developing children (see Engberg-Pedersen
and Thomsen (2016)). The JDV test is intriguing: it bears a clear family resemblance
to second-order false belief tests, and in our study we used it alongside four standard
second-order false belief tasks. Nonetheless, as we shall see, the link between second-
order false belief reasoning and JDV reasoning is not straightforward. In particular,
we found that sentential complementation skills were not significantly correlated
with JDV scores. As second-order false belief reasoning and JDV reasoning seem
very similar, this calls for further explanation. The aim of this paper is to get a better
understanding of the issues involved.
We proceed as follows. In Sect. 2, we introduce second-order false belief tasks
and their role in the study of Theory of Mind. In Sect. 3, we discuss the concept of
recursion, its role in linguistics, and why it may be relevant to children with ASD.
In Sect. 4, we describe our experimental setup, and in Sect. 5 we present the results
which show that mastery of linguistic recursion is a significant predictor of success
in second-order false belief tasks for children with ASD. With these preliminaries
behind us, we reach the main theme of the paper: JDV reasoning and how it relates to
second-order false belief reasoning. So in Sect. 6, we backtrack and discuss (with the
help of examples) JDV reasoning and the kind of logic it seems to involve. Then, in
Sect. 7, we present our experimental results; as we shall see, the mastery of sentential
complementation is not significantly correlated with JDV. In Sect. 8, we try to gain
a better understanding of what second-order false belief and JDV reasoning do (and
do not) have in common, and what this may tell us about how children with ASD
acquire their mentalizing skills.
2 Second-Order False Belief Tasks
According to the fifth edition of the American Psychiatric Association’s Diagnostic

and Statistical Manual of Mental Disorders (DSM-5), ASD is a neurodevelopmental
disorder characterized by impairments in social behavior and communication, as
well as by restricted, repetitive behavior. As we shall discuss in Sect. 5, the mastery
of linguistic recursion (and in particular, sentential complementation) is a significant
predictor of success in second-order false belief reasoning in children with ASD. In
this section, we introduce second-order false belief reasoning by presenting one of
the four standard second-order false belief tasks used in our experimental work. This
Linguistic Recursion and Danish Discourse Particles: Language … 23
will lead us to a discussion of the role played by false belief tasks (both first-and
second-order) in the study of Theory of Mind (ToM), and to the ToM deficit account
of ASD.
The task we will present is the second-order Sally–Anne task (Baron-Cohen et al.
1999). It is the shortest of the four second-order false belief tasks we used, and the
one that most clearly illustrates the distinction between first- and second-order tasks.
A child is shown a scene with two doll protagonists, Sally and Anne, with a basket and a box
respectively. Sally first places a marble into her basket. Then Sally leaves the scene, and in
her absence, Anne moves the marble and puts it in her box. However, although Anne does
not realise this, Sally is peeking through the keyhole and sees what Anne is doing. Then
Sally returns, and the child is asked: “Where does Anne think that [Sally will] look for her
marble?”
Typically developing children aged six or more usually handle second-order false
belief tasks correctly (see Miller (2009, 2012)). They reply that Anne thinks that
Sally will look in the basket, which is where Anne (wrongly) believes that Sally
believes the marble to be. Most younger children reply that Anne thinks that Sally
will look in the box. This is indeed where Sally knows the marble to be—but Anne
does not know that Sally knows this, so this response is incorrect. In short: in order
to pass the test, the child must ascribe a false belief to Anne. Moreover, note that it
is a belief about Sally’s belief, that is, an embedded belief.
For this reason, the task is called the second-order Sally–Anne task: it deals with
beliefs about beliefs as opposed to beliefs about ordinary facts about the world. To
make this more concrete, let us compare this second-order test with the first-order
Sally–Anne task (Baron-Cohen et al. 1985). The (better known) first-order version
of the test can be obtained from the second-order task simply by deleting the material
in bold font and switching [Sally will] to “will Sally”:
A child is shown a scene with two doll protagonists, Sally and Anne, with a basket and a
box respectively. Sally first places a marble into her basket. Then Sally leaves the scene, and
in her absence, Anne moves the marble and puts it in her box. Then Sally returns, and the
child is asked: “Where will Sally look for her marble?”
In the second-order task, the child was asked about Anne’s belief about Sally’s belief
about the location of the marble (a second-order belief), whereas in the simpler first-
order task, the child is just asked about Sally’s belief about the location of the marble
(a first-order belief). Nonetheless, note that—just as in the second-order test—to
answer correctly the child must ascribe a false belief, though this time to Sally.
Extensive experimental work with first-order false belief tasks has shown the
existence of a transition age, but it is lower than in the second-order case: children
above the age of four will usually say that Sally will look in the basket, which is
where Sally (falsely) believes the marble to be. Children younger than four, however,
will usually say that Sally will look in the box: this is where the marble is, but Sally
does not know this, and hence the answer is incorrect. In short, understanding how to
make use of first-order beliefs is a cognitive skill that children pick up earlier than the
skill involved in handling second-order beliefs. Moreover, they are developmentally
distinct stages for both typically developing children and children with ASD.
But distinct stages in the development of what? A standard answer here is Theory
of Mind (ToM). This is a broad term and has a number of synonyms or near-synonyms
including mindreading, empathy, folk psychology, social cognition, and taking the
intentional stance. Roughly speaking, ToM is the ability to infer and attribute mental
states in order to explain and predict behavior. It is considered to be an important
human ability, one that adults typically apply effortlessly and in a wide range of
situations, such as when they help someone who looks lost without needing to be
asked, or understand an ironic remark. Human beings are social creatures, and the
acquisition of ToM is part of what enables them to enter the web of human interaction.
ToM first arose in primatology (see Premack and Woodruff (1978)) but it is now a
central concept in cognitive and developmental psychology. Moreover, ToM and false
belief tasks are deeply interlinked. Indeed, false belief tasks were devised precisely
because they offered a plausible empirical handle on the interesting (but elusive)
notion that humans (and other primates) possess a “Theory of Mind,” and, starting
with the classic work of Wimmer and Perner in the 1980s on both first-order tasks
(see Wimmer and Perner (1983)) and second-order tasks (see Perner and Wimmer
(1985)) the false belief handle has proved fruitful: it has generated an impressive
body of empirical data about when ToM is acquired and what factors facilitate its
development (see Wellman et al. (2001) for an influential meta-analysis).
False belief tasks also fuelled the ToM deficit account of ASD. This account stems
from a heavily cited 1985 study by Baron-Cohen and his colleagues (see Baron-
Cohen et al. (1985)) which used the first-order Sally–Anne task to demonstrate a
difference in ToM profiles for typically developing children and children with ASD;
since then, the idea that delayed or different development of the ability to represent
other minds is a root cause of ASD has been highly influential (see Tager-Flusberg
(2007)). ToM limitations, it is argued, may give rise to misreadings of cues from
others, to the misinterpretation of social situations, and to limited expressions of
empathy and reciprocity. Even though some studies report findings that do not support
the ToM deficit account (see Bowler et al. (2005)), and other significant theoretical
accounts have been put forward (see Rajendran and Mitchell (2007)), the ToM deficit
account remains influential and is currently applied in both theoretical and clinical
work.
To briefly recap what we have said: attaining false belief competency is usually
associated with the acquisition of ToM. Moreover, false belief competency comes in
two distinct stages for both typically developing children and children with ASD: the
attainment of first-order and second-order belief reasoning competencies. However,
although both competencies are viewed as important stages of ToM development,
a lot more is known about the first-order stage than the second. Over the past 35
years, both correlational and training studies (involving both typically developing
children and children with ASD) have yielded rich and robust first-order results that
cover a wide range of task manipulations and experimental setups, and it is now well
established that children’s understanding of other minds depends on various language
skills and on working memory development. However, much less is known about
second-order reasoning, and many conclusions about it are tentative (Miller 2009,
2012), even for typically developing children. The literature examining second-order
false belief reasoning in children with ASD is even smaller, which was one of the
motivations for our study.
3 Recursion
The central hypothesis of our correlation and training study (Polyanskaya 2018) was
that linguistic recursion skills contribute to the development of second-order false
belief competency in children with ASD. The purpose of this section is to motivate
the hypothesis.
The best known examples of recursion in linguistics are embeddings of phrases
inside phrases of the same type. Such embeddings allow hierarchical structures of
arbitrary complexity to be formed. Embeddings can be thought of either structurally
(declaratively) or algorithmically (procedurally). Both perspectives are useful, indeed
they are two sides of the same coin, and linguists use both. In syntax textbooks, phrase
structure rules (such as S −→ S or S) are often introduced procedurally: this one
would be viewed as a rule for producing (or parsing) disjunctive sentences. But it
can also be viewed structurally: it says that a tree node labeled S is “licensed” if it
dominates precisely three nodes labeled S, or, and S, respectively.
The dual perspective on recursion is also commonplace in computer science,
where one talks about both recursive data structures and recursive algorithms. Again,
the two perspectives are linked. Recursive data structures, such as lists or trees, can
be assembled (or dismantled) using recursive algorithms; conversely, a recursive
algorithm for solving (say) the Tower of Hanoi problem will typically do so by
recursively manipulating the data structures representing the problem.1
Recursion plays an essential role in contemporary syntax and semantics. Indeed,
these two levels work together: the recursive structural analyses provided by the syn-
tax guides the semantic construction process. The keyword here is compositionality:
the meaning of whole is built from the meaning of the parts, with syntax telling us
what the parts are.2 The partnership of syntax and semantics is clear in embedded
complements:
I think [ S Bill hopes that [ S Susan wishes that [ S Fergus will mess up at next week’s meeting
]S ]S ]S .
1 The Tower of Hanoi problem is on how to transfer a stack of disks, one disk at a time, from one
location to another, with the help of a third (intermediate) location. The disks are all of different
sizes, and the key rule is that a bigger disk may never be placed on a smaller one. The problem can
be solved elegantly (for any number of disks) by a simple recursive program. YouTube has many
good videos illustrating how this works.
2 For a more detailed discussion of compositionality and the structure of the syntax-semantics
interface, see Szabó (2013).

The brackets mark the recursive embeddings of the sentential complements. The
sentence’s meaning (that is, what it is that I think) is compositionally formed from
the meanings of its embedded complements.
Why might linguistic recursion be relevant to second-order false belief tasks and
ASD? It is relevant to second-order false belief tasks because it is linguistic recursion
that enables us to build sentences to talk about beliefs about beliefs (Anne believes
that Sally believes that the ball is in the basket) and indeed, about beliefs about
beliefs about beliefs, and so on. It may be particularly relevant to ASD because of
the possibility it offers of “bootstrapping” false belief understanding using language
skills. Here’s the way Tager-Flusberg and Joseph put it:
Our research suggests that one way in which language plays a crucial role in these outcomes
is that it helps to bootstrap an understanding of false beliefs and related aspects of social-
cognitive components of theory of mind. Some children with autism, the minority with
normal or near-normal linguistic ability, can use language to reason logically through false
belief tasks, or to interpret what others know or believe on the basis of their experience with
specific events. (Tager-Flusberg and Joseph 2005)[page 311]
To spell out the underlying ideas somewhat: typically developing children develop
their mentalizing skills via social interaction, but this developmental path is problem-
atic for children with ASD due to their impaired social skills. In the above quotation,
Tager-Flusberg and Joseph note that linguistic skills may offer an alternative devel-
opmental path which compensates for this. Moreover, they claim that knowledge
of sentential complementation may be particularly relevant here. As they put in the
abstract to the paper:
Children with autism are especially dependent on language, particularly knowledge of sen-
tential complements, to bootstrap their meta-representational capacity. (Tager-Flusberg and
Joseph 2005)[page 298]
Tager-Flusberg and Joseph only discuss first-order reasoning. We hypothesized that

linguistic recursion (and in particular, sentential complementation) might provide a
“bootstrapping” tool of relevance to second-order tasks. Our own previous logical
analyses of second-order false belief tasks (see Braüner et al. (2016a, b)) highlighted
the importance of recursion: it showed that second-order reasoning can be viewed as
the recursive embedding of first-order reasoning about different agents. Moreover,
several studies have commented on the developmental distinction between first- and
second-order phenomena. To quote one example:
We are convinced that there is a difference between first- and second-order embedding phe-
nomena, both in language and cognition. The core of this must lie in the exclusive explicitness
that a system of recursion provides, where discourse representations allow too many or too
few represented connections to enable deductive reasoning. (Hollebrandse 2008)[page 276]
It thus seemed plausible that linguistic recursion might play an important “bootstrap-
ping” role in the development of second-order skills.3
3 As far as we are aware, the relation between recursive embedding in language and second-order
false belief reasoning has been the focus of only a handful of previous studies (see Arslan et al.
One final remark: as logics capable of handling recursive embeddings of beliefs

are more computationally complex than those required to analyze first-order tasks,
processing issues are relevant and should be experimentally investigated. A recent
study has identified working memory as a significant predictor of second-order false
belief competencies in the typically developing population (see Arslan et al. (2017));
the results we discuss in Sect. 5 show that it is relevant to the ASD population as
well.
4 Experimental Setup
In this section, we describe the correlation component of our study (Polyanskaya

2018; Polyanskaya et al. 2018). We begin with our subjects and how they were
recruited. As we have mentioned, there were 62 children in our sample. Their age
range was 6–16 years old. The four most frequent age groups were: 10-year-olds
(19%), 11-year-olds (16%), 12-year-olds (11%), and 13-year-olds (14.5%). Most
participants were recruited from schools for children with special educational needs
in the Sjælland region of Denmark (which contains the capital city Copenhagen and
rural areas) though five were recruited via direct contact with parents. The children
had to satisfy the following recruitment criteria: parental consent had to be obtained,
they had to be diagnosed with ASD based on a formal evaluation by a specialist, they
had to be aged 6–16, and they could not be undergoing medical treatment affecting
cognitive performance. Moreover, they had to be monolingual Danish native speakers
with no learning difficulties or language delays at the time of recruitment. Finally,
they had to have the emotional readiness to undergo a testing situation (and a training
program); this last decision was made on the basis of teacher assessment.
We started by pre-testing. Children were given standard tasks from verbal com-
prehension and working memory indices (WISC-IV) as well as general grammar
comprehension (TROG-2). These results served as quantitative inclusion criteria:
we only selected children with working memory and language skills within the age
norms specified by these tests. The standard score for verbal comprehension (VC)
and working memory (WM) had to be at least 80 for a child to be included in the
study.
Our correlation study then used the 62 children who met the pre-testing criteria:
they were given second-order false belief tasks and tested for their competence with
sentential complements. In addition, we also gave 49 of these children the JDV test,
a gap-filling test of Danish dialogue particle skills. As the second half of this paper
is devoted to a detailed discussion of the JDV test, we won’t say more about it here,
except to remark that it is a written test, and only 49 children in our sample had the
(2017), de Villiers et al. (2014), Hollebrandse (2008), Bogaerds-Hazenberg and Hendriks (2016)).
All these studies investigate typically developing children, and to the best of our knowledge, ours is
the first to experimentally investigate the relationship between recursion in language and recursive
reasoning for children with ASD.
reading/writing skills required to take it. In addition, we asked teachers to complete

a Social Responsiveness Scale (SRS) questionnaire for all 62 subjects. The SRS is
the validated quantitative measure of autistic traits developed by Constantino et al.
(see Constantino et al. (2003)).
The four second-order false belief tests we used were Danish translations of
standard second-order false belief scenarios: the ice-cream task (see Perner and
Wimmer (1985)), the puppy task (see Sullivan et al. (1994)), the (second-order)
Sally Anne (see Baron-Cohen et al. (1999)), and the bake sale task (see Hollebrandse
et al. (2014)). Two of the tasks (ice-cream and puppy) were taken from the Danish
version of the validated Dutch ToM Storybook Frank (see Blijd-Hoogewys et al.
(2008)), translated and validated for Danish by Clemmensen et al. (see Clemmensen
et al. (2016)); for a detailed discussion of all the second-order test material used,
see (Polyanskaya 2018; Polyanskaya et al. 2018). The four second-order false belief
tasks were randomized in the order of presentation across the children.
As far as we are aware, these four tests are representative of the standardly used
second-order false belief tests; all other second-order false belief tests we know of
are variants of one of the four we used. Moreover, our previous logical analyses
of second-order false belief reasoning had enabled us to classify them across two
dimensions of binary variation (see Braüner et al. (2020)) and to show that the four
we selected covered all four possible variants. So our second-order false belief test
battery was comprehensive.
For present purposes, however, we can ignore the differences between the tests.
What is important here is that in all four of them, obtaining the correct answers
requires the child to understand what one character in the story falsely believes about
another character’s belief. We’ve already seen an example of this in the second-order
Sally–Anne task, but to gain a clearer impression of what is involved, let’s look at
the puppy story as well:
Tonight it’s Peter’s birthday and Mom is surprising him with a puppy. She has hidden the
puppy in the basement. Peter says, “Mom, I really hope you get me a puppy for my birthday.”
Remember, Mom wants to surprise Peter with a puppy. So, instead of telling Peter she got
him a puppy, Mom says, “Sorry Peter, I did not get you a puppy for your birthday. I got you
a really great toy instead.”
Now, Peter says to Mom, “I’m going outside to play.” On his way outside, Peter goes down
to the basement to fetch his roller skates. In the basement, Peter finds the birthday puppy!
Peter says to himself, “Wow, Mom didn’t get me a toy, she really got me a puppy for my
birthday.” Mom does not see Peter go down to the basement and find the birthday puppy.
Now, the telephone rings, ding-a-ling! Peter’s grandmother calls to find out what time the
birthday party is. Grandma asks Mom on the phone, “Does Peter know what you really got
him for his birthday?” Now remember, Mom does not know that Peter saw what she got him
for his birthday. Then, Grandma says to Mom, “What does Peter think you got him for his
birthday?”
Second-order false belief question: What does Mom say to Grandma?
Crucially, none of the test questions in any of the four stories we used contained any
recursive sentential complements. That is, participants did not need to have mastered
this form of linguistic recursion to understand the questions. Moreover, children
were asked memory or first-order false belief questions to check that they really had
understood the scenarios. In total, children were asked 10 second-order false belief
questions (6 judgment and 4 justification) and 11 control questions (4 first-order and
7 memory). The score range for judgment answers was 0–6 (two of the tests have
2 judgment questions) while for justifications it was 0–12, yielding our composite
score range of 0–18. A justification score of 0 was given for incorrect explanation.
A justification score of 1 was given for correct reasoning without any references to
mental states (Because Mom said she gave him a toy), a score of 2 was given for
correct reasoning using one mental state verb (Because Mom does not know he saw
the puppy), and 3 was given for using two or more mental verbs (Because Mom thinks
that he does not know that she brought a puppy). Participants were only credited with
points for second-order false belief questions if they had passed the control question.
To conclude this section, we turn to the test for sentential complement skills.
We found no validated test in Danish that measured the ability to comprehend
(or produce) any recursive syntactic construction, so we developed the Recursive
Embedding Tool (RET). The pilot version was validated on 240 typically developing
Danish-speaking children, and the final version on 70 typically developing Danish-
speaking children (36 girls and 34 boys) and 15 adults. The final version of RET
initially consisted of two subsections, one utilizing possessive noun phrases and one
on sentential complements, but since there was a clear ceiling effect for the posses-
sive nouns subsection, we only used the sentential complements subsection in the
present study.
The items for the subsection on sentential complements were shaped as illus-
trated short stories concluded by questions. The stories and accompanying images
were inspired by the hypothesis (see de Villiers et al. (2014)) that to recognize gen-
uine sentential complement recursion, children need truth-value contrasts between
clauses. Thus, the stories include a statement that does not correspond to the story-
world’s reality in order to prompt the recursive reading. This motivates the RET
example of Fig. 1: note the truth-value contrast between the rain outside the window
and Mom’s statement that the sun is shining. Inspired by Hollebrandse and Van Hout
(2015), each item involved both a single-embedding and a double-embedded ques-
tion. Participants received 1 point for a correct answer to the double embedding
question, and 0 otherwise.
A last remark: as we mentioned earlier, as well as the correlation study, we also
carried out a training study. The idea was to train participants to grasp the following
four principles:
1. That several linguistic constituents of the same type may be combined together.
2. That these constituents may be embedded one inside another.
3. That changing the order of embedding changes the meaning.
4. That the number of embedded constituents is potentially unlimited.
Fig. 1 A RET example
That is, in effect, we taught them basic ideas of compositional semantics; the training
materials used the RET format of illustrated stories. We won’t discuss this aspect
of our work here; the interested reader should consult Irina Polyanskaya’s Ph.D.
thesis (Polyanskaya 2018).
5 RET and Second-Order False Belief Reasoning
As we shall now see, RET is a significant predictor of success in second-order false

belief tasks for children with ASD. However, as the main goal of this paper is to
discuss why RET was not significantly correlated with JDV, we will only give the
basic correlation and regression results. For a more detailed discussion, we refer the
reader to Polyanskaya et al. (2018).
We began by inspecting the interrelationships among composite second-order
false belief scores, recursive complements (RET), general grammar comprehension
(TROG), working memory (WM), age in years, autistic severity symptoms (SRS),
and verbal comprehension (VC). As not all variables were normally distributed, we
used Kendall’s τb correlation coefficient, a nonparametric measure of the strength
and direction of association. Table 1 shows the correlation coefficients and strength
among study variables (we have abbreviated “second-order false belief” to SOFB).
Table 1 Kendall’s τb correlations (Sig. Levels: * p <0.05, ** p <0.01)

Variables SOFB RET TROG WM Age SRS VC
SOFB 1
RET 0.28∗∗ 1
TROG 0.27∗∗ 0.18 1
WM 0.36∗∗ 0.19 0.30∗∗ 1
Age 0.34∗∗ 0.09 0.30∗∗ 0.40∗∗ 1
SRS –0.16 –0.03 0.00 0.00 –0.18 1
VC 0.43∗∗ 0.24∗ 0.46∗∗ 0.46∗∗ 0.56∗∗ –0.24∗ 1
Table 2 Regression analysis

Variable B SE t p
Age 0.46 0.24 1.91 0.07
WM 0.22 0.13 1.69 0.09
TROG 0.23 0.35 0.66 0.50
RET 0.63 0.31 2.03 0.04
Second-order false belief scores correlated significantly and positively with all vari-
ables except SRS (where, as expected due its scoring design, there was a negative
correlation, but one that was not significant).
Moreover, hierarchical regression analysis showed that RET predicts second-order
false belief understanding, when controlled for age, WM, and TROG. The second-
order false belief score was the dependent variable and age, working memory, and
TROG were entered first, followed by RET scores. All relevant statistical assump-
tions were checked. With age, WM, and TROG, the model explained 29% of the vari-
ance in second-order false belief understanding, F(3, 54) = 7.31, p < 0.001. With
RET scores, the model explained an additional 5% of the variance, F(4, 53) = 6.88,
p < 0.001. Regression coefficients, standard errors, and p-values for the full model
are presented in Table 2. This table shows that RET is a significant predictor, that
age and working memory are marginally significant predictors, but that grammar
comprehension is not significant.
More could be said here. For example, we also divided the total working memory
score up into simple and complex memory scores—the difference being that simple
memory tasks tap what is usually referred to as short-term memory, while complex
memory tasks involve the ability to manipulate information kept in short-term mem-
ory. In the subsequent regression analysis, the complex WM variable turned out to be
a significant predictor of second-order false belief, together with age and RET, but
simple WM was not. To put it another way: splitting the marginally significant total
WM variable indicated the predictive role of complex WM tasks (see Polyanskaya
et al. (2018)).
We view these results (together with the training results and comparisons with typ-
ically developing children reported in Polyanskaya (2018)) as offering partial support
for the view that competency in linguistic recursion may act as a “bootstrapping”
strategy, or a “scaffold”, that enables children with ASD to acquire second-order
false belief competency.
6 JDV Reasoning
We are now ready to consider a somewhat puzzling result. As we have already

mentioned, alongside our second-order false belief test battery we also administered
the JDV test. As we shall see, the sort of reasoning required for JDV success seems
very close to second-order false belief reasoning. However, RET was not significantly
correlated with JDV. In this section and the next, we shall explain the issues involved,
and try to make sense of them.
The JDV test (Thomsen 2012; Engberg-Pedersen and Thomsen 2012) is a Dan-
ish gap-filling task of competence with the Danish discourse particles jo, da, and
vel. The test was designed as a language test requiring social cognition skills,
and was successfully employed by its creators in a matching study between Danish-
speaking children with ASD and typically developing children (Engberg-Pedersen
and Thomsen 2016).
Danish has a group of dialogue particles (there are said to be nine of them)
which allow speakers to signal the relationship between their own understanding of
a situation, and how their addressee understands it. They are not simple particles to
master. As the creators of the JDV test put it (Engberg-Pedersen and Thomsen 2016,
p. 1):
Acquiring dialogue particles requires sophisticated perspective-taking skills as children must
be able to entertain a state of affairs taking into account both their own mental state and
another person’s mental state simultaneously. Furthermore, the acquisition of dialogue par-
ticles may be especially taxing because of their perceptual inconspicuousness: the particles
are unstressed in non-salient, utterance-medial positions.
Let’s look at the three relevant particles in turn, starting with jo, the simplest. Jo
is used to indicate that the speaker expects the addressee’s agreement. To make
matters concrete, suppose Torben and Patrick (both modal logicians) together attend
an inspirational lecture given by a famous modal logician called Johan. Toward the
end of the lecture, Torben turns to Patrick and whispers:
Det var jo et perfekt foredrag.
Had Torben merely said Det var et perfekt foredrag—that is, had Torben omitted the
jo—the English translation would simply be That was a perfect lecture. But the jo
adds something extra: based on his knowledge of the common ground (for example,
the quality of the lecture, and his knowledge of Patrick’s knowledge of modal logic),
Torben is not only affirming that it was a perfect lecture, but he is also signaling that
he expects Patrick’s agreement.
It will be helpful for our later discussion if we develop some impression of the
kinds of logical tools required to model jo, da, and vel. Let’s try starting with propo-
sitional modal logic. Suppose modality Bs means “the speaker believes,” Ba means
“the addressee believes,” and p is shorthand for the (jo-free) sentence Det var et per-
fekt foredrag. Then (as a first approximation) we might say that Torben’s utterance
of Det var jo et perfekt foredrag means something like the following:
p ∧ Bs p ∧ Bs Ba p.
That is, the effect of enriching an assertion of p with jo is to explicitly add the
information that the speaker believes that p (that is, Bs p) and also believes that the
addressee does too (that is, Bs Ba p). Of course, it might be argued that here we really
need knowledge operators K a and K s . But for present purposes, all that is relevant
is that we we can spell out, reasonably plausibly, at least some of the semantics of
jo in a standard (static) propositional modal logic. But with vel and da, matters are
not so straightforward.
So let’s turn to vel. Suppose Torben, Irina, and Patrick agreed some weeks ago to
meet and discuss second-order false belief reasoning. The day before the meeting,
Irina phones Torben (to discuss a statistical problem) and in the course of their
discussion, Torben mentions that he has to give a rather demanding lecture for post-
graduate students the very next day. Irina responds:
Du kommer vel til mødet?
Had Irina merely asked Du kommer til mødet?, the English translation Are you
coming to the meeting? would suffice. But the vel adds something new: based on her
knowledge of the common ground (for example, the fact that their appointment was
made a long time ago, and that preparing for a difficult lecture takes a lot of time),
Irina in addition signals expectation and uncertainty—she is uncertain as to whether
Torben will show up tomorrow, but appeals to Torben to confirm her expectation that
he will. To put it another way: she admits her uncertainty, and concedes that Torben
has privileged access to the information relevant to resolving it.
Clearly, vel is more subtle than jo, but we might attempt to capture its import
as follows. Suppose we use the operator (?s ) to mean “the speaker is uncertain,”
the operator (?a ) to mean “the addressee is uncertain,” and use q as shorthand for
the (vel-free) sentence Du kommer til mødet. Then we might try representing Irina’s
utterance of Du kommer vel til mødet? by
q ∧ (?s )q ∧ Bs ¬(?a )q.
That is, to a first approximation, the effect of enriching an assertion of q with vel
is to explicitly add that the speaker is uncertain about q (that is, (?s )q) and also to
concede that the addressee is not uncertain (that is, Bs ¬(?a )q). But here we enter the
less standard logical territory. Even if we put aside the increasingly dynamical flavor
(Irina will have to change her mind about tomorrow if Torben decides not to show
up), we are still left with the task of devising a suitable formal semantics for (?s ) and
(?a ), one that interacts correctly with standard belief (or knowledge) operators; this
is an interesting task, but not one we shall attempt here.
And now for da. This is the most subtle of the three, and we’ll look at two examples.
Suppose a friend of yours turns up wearing a new shirt. It’s very colorful and highly
unconventional—but you think it looks great! However, you can tell from the look
on your friend’s face that she thinks she has made a mistake, or is at least uncertain
about it. But you have no doubts and say:
Det er da en flot skjorte.
That is, instead of simply saying Det er en flot skjorte, which means That’s a nice
shirt, you added the da. This simultaneously signals (a) your certainty that you
are right, (b) that there is possibly a conflict of perspectives involved (between the
speaker’s perspective and the addressee’s), and yet (c) it also seems to affirm that the
correct understanding of the situation is already known to the addressee.
For our second example, let’s return to the Torben–Irina–Patrick meeting. Suppose
that just after her phone conversation with Torben, Irina phones Patrick, and the topic
of Torben’s lecture comes up. Patrick (who has just spent the whole day making notes
for the meeting) responds:
Han kommer da til mødet.
Had Patrick said Han kommer til mødet, the English translation would simply be He
is coming to the meeting, but his addition of da expresses (a) his certainty that Torben
is coming, (b) signals the presence of conflicting information, and (c) affirms that
Irina (deep down) knows that Torben really has to show up for that meeting.
Let’s try capturing (some of) this in symbols. Let r be shorthand for the (da-free)
sentence Han kommer til mødet. Then we could try representing part of the meaning
of Patrick’s utterance of Han kommer da til mødet by
r ∧ Bs r ∧ Bs (Ba ¬r ∨ (?a )r ).
This (static) formula says that the effect of enriching an assertion of r with da is
(a) to explicitly add that the speaker believes r (that is, Bs r ) and (b) to affirm that
the speaker believes that either the addressee believes that r is false (that is, Ba ¬r )
or the addressee is uncertain about r (that is, (?a )r ). But, at best, this only captures
parts (a) and (b) of Patrick’s utterance (or, for that matter, only parts (a) and (b)
of our first example). Part (c) of Patrick’s utterance (and of our first example) has
dynamic effects that are hard to ignore. There is some sort of insistence that Irina must
change her mind. It’s almost as if you can hear a but in the background: something
like But you know Torben’s got to turn up, right? can be felt. Clearly, someone’s
information needs updating, and if we want to do some logical modeling here it would
be wise to start using ideas from dynamic frameworks such as Dynamic Epistemic
Logic (Van Ditmarsch et al. 2007) or Discourse Representation Theory (Kamp et al.
2011). But, once again, we won’t turn to the logical modeling task here; instead,
we’ll turn to what we know experimentally.
7 The JDV Test
As should be clear from our introductory examples, jo, da, and vel are semantically
subtle. But what is important for the present paper is not the precise meaning of these
particles, but the fact that their meanings are stable enough to support an accurate test:
the JDV test. As we mentioned earlier, this test is a gap-filling test. The participants
read small stories and fill in the gap with one of these particles. Here’s an example
(taken from Engberg-Pedersen and Thomsen 2016, p. 6):
Signe has just been to Mia’s birthday.
Signe asks Mia: Wasn’t it a nice present we had found for your birthday?
Mia answers:
Helt klart! Jeg elsker tegneserier
Mia’s response means Definitely! I love comics. Many of the JDV test items come
with simple illustrations, and characters recur, so the child answering this has met
Signe and Mia before. And the question is: how should this gap be filled? With jo,
da, or vel? The answer is jo. Here’s why (Engberg-Pedersen and Thomsen 2016,
p. 7):
This is a prototypical context for jo because it presents shared knowledge and agreement
between the speaker and the addressee, who have been introduced as close friends in pre-
vious items, and who can therefore be expected to be aware of each other’s taste. There
is no opposing viewpoint to motivate da (shared knowledge and conflict), and the speaker
obviously knows more about her own taste than the addressee, thus excluding vel (agreement
and privileged recipient knowledge).
The JDV test consists of 15 such items, scored 1 for the correct answer, 0 otherwise.
Adult Danish native speakers exhibit a high degree of agreement on the answers:
all 60 adults in the pilot test scored between 13 and 15. When tried out on typically
developing children (Danish native speakers), the results indicated that 11–14-year-
olds followed adult-like usage norms (see Engberg-Pedersen and Thomsen 2016, p.
7). The creators of the JDV test are well aware of the resemblance between their test
and second-order false belief tests (Engberg-Pedersen and Thomsen (2016, p. 8));
indeed, they say that they are equivalent:
The JDV-test is even more demanding than most false belief tasks as it requires a different
type of mental state understanding than the traditional so-called first-order false belief tasks.
In order to choose the most appropriate dialogue particles, the children need to understand
a story character’s understanding of a different story character’s perspective. This task is
equivalent to so-called second-order false belief tasks, tasks that typically developing chil-
dren pass around the age of six.
They used the test on children with ASD, their key result being that children with
ASD score lower on the JDV test than typically developing children. To be precise:
the mean JDV score of a group of children with ASD (n = 28) was significantly lower
than the mean score of a group of typically developing children (n = 29), matched
on chronological and nonverbal mental age as well as vocabulary and grammar
comprehension.
Fig. 2 JDV score histogram
We also found the JDV test intriguing (for much the same reasons as its creators
did) and included it in the pre-testing phase of our work so that we could run corre-
lations and regressions on it; 49 of our test subjects took the JDV test.4 The mean of
our JDV scores was rather high, 10.94 (SD= 3.28). Figure 2 is a histogram of our
JDV scores. It indicates a non-normal distribution with two clusters around 9 and 14
points.
As the distribution was non-normal, we used Kendall’s τb correlation coefficient.
We found that JDV correlates significantly with grammar comprehension (0.39,
p<0.01) and verbal comprehension (0.24, p<0.05), thus, our data supports the cor-
relations with JDV mentioned in Engberg-Pedersen and Thomsen (2016).5 More
4 As we mentioned earlier, the main reason that not all of our 62 subjects took the JDV test is that
it is a written test; not all our subjects had the required reading/writing skills to take it.
5 The focus of Engberg-Pedersen and Thomsen (2016) is the matching study, and they only mention
in passing a few exclusively ASD results, namely positive correlations for JDV with grammar com-
prehension, with verbal comprehension, and with nonverbal cognitive abilities. They also mention
that: “The children were also given two false belief tests, one of location change and one of unex-
pected contents with altogether three questions about beliefs,” but no statistical results are given
(Engberg-Pedersen and Thomsen (2016, p. 9)).
importantly for the purpose of the present study, however, there are also significant
correlations with second-order false belief tasks (0.27, p<0.05), working memory
(0.36, p<0.01), and age (0.21, p<0.05). However, no significant correlation with RET
or SRS was found. We don’t find the lack of significant correlation between JDV
and SRS particularly interesting,6 but the lack of a significant correlation between
JDV and RET calls for further explanation; we will turn to this task in the following
section.
Before doing this, we’ll note our other JDV-related results. In order to investigate
the contribution of discourse particle skills to second-order false belief competency,
after controlling for age, working memory, grammar, and RET, a hierarchical regres-
sion analysis was conducted. In the first step, the model included all the variables
that we wanted to control (the variables in Table 2); and in the second step, the
main variable of interest, JDV, was added to the model. The statistical assumptions
were checked. The results showed that discourse particle competency did not add
significantly to the model. The model without JDV accounted for 34% of the vari-
ance in second-order false belief understanding, and with JDV, the model explained
only an additional 1% (F(5, 41) = 4.50, p = 0.002), but the change was not sig-
nificant ( p = 0.37). The individual beta weights for the full model were as follows:
age B = 0.47, p = 0.10; working memory B = 0.16, p = 0.30; TROG B = 0.07,
p = 0.86; RET B = 0.65, p = 0.06; JDV B = 0.19, p = 0.37.
Given the number of variables involved, it is probably unsurprising that JDV
seems to contribute so little. However, we decided to test the possibility of gram-
matical knowledge and false belief skills being predictors of JDV competency (we
read the creators of the task as hypothesizing this in Engberg-Pedersen and Thomsen
(2016)). Accordingly, a multiple regression analysis was conducted with JDV as a
dependent variable, and age, second-order false belief (SOFB), raw TROG, and raw
WM as independent variables. The sample size allowed us to include four indepen-
dent variables; RET was not included as it was not correlated with JDV. Missing
values were excluded listwise, and thus the sample included 49 participants. Before
interpreting results, statistical assumptions for multiple regression were checked.
The multiple regression model statistically significantly predicted JDV scores,
F(4, 41) = 9.93, p < 0.001, R 2 = 0.49 (Cohen’s f 2 = 0.96; large effect). The indi-
vidual beta weights were as follows: age B = 0.20, p = 0.29; working memory
B = .33, p = 0.005; TROG B = 0.85, p = 0.002; SOFB B = 0.002, p = 0.98,
showing that age and SOFB did not significantly predict JDV, while TROG and
working memory were significant predictors.
As noted above, we found significant correlations between JDV and grammar
comprehension (TROG) as well as between JDV and working memory. Moreover, the
regression analysis just given shows that grammar and working memory both predict
JDV. These results are in line with results from Engberg-Pedersen and Thomsen
(2016), which reports that for children with ASD, better grammar comprehension,
and nonverbal mental ability were associated with a better comprehension of the
6 Note that SRS does not correlate significantly with the other variables either, and that SRS is rather
different from the other variables in that SRS is filled out by the children’s teachers.
dialogue particles. Even though a direct comparison between our working memory
scores and their nonverbal measure is not possible (as they applied the Matrices
subtest of the Wechsler Nonverbal Scale of Ability), it is noteworthy that JDV is
associated with grammar and nonverbal measures in all analyses.
8 Concluding Discussion
Second-order false belief and JDV reasoning clearly have a lot in common. It is
easy to see why the creators of the JDV test described it as “equivalent to so-called
second-order false belief tasks”. Both types of tasks involve reasoning about the
perspective of different agents, and the fact that we found significant correlations
with second-order false belief tasks and working memory may suggest that there is
nothing more to say here. But we find the lack of significant correlation with RET
interesting. Is there more that can be said about this (somewhat puzzling) result? We
think there is.
We begin by making three general points. First, for all their similarity, JDV reason-
ing and second-order false belief are logically rather different. The style of reasoning
used in second-order false belief tests can be formalized in relatively simple static
propositional modal languages, and the proof structures underlying it have a simple
and elegant recursive structure (support for both these claims is spelt out in detail in
Braüner et al. (2016b)). JDV reasoning, on the other hand, is less straightforward.
Indeed, as our brief remarks on formalization in Sect. 6 should have made clear, trying
to get to grips with it technically swiftly leads to a logically novel (and interesting)
territory.
Second, JDV reasoning is not universal in the way second-order false belief rea-
soning is. Discourse particles that rely on perspective taking can be found in other
languages, but they are hardly systematic. JDV reasoning is an interesting collec-
tion of reasoning patterns about perspectives, but it does not exhibit a pattern that
underlies even all other Germanic languages (English has no jo counterpart). Second-
order false belief reasoning, on the other hand, is something that is common to many
(arguably all) cultures. It is not intrinsically linguistic (let alone grammaticized) in
the way that JDV reasoning is. Rather, it is a general form of reasoning based around
a pervasive concept called (in English) ‘belief’.
And this leads to the third point. What we mean by second-order false belief
reasoning is actually the tip of a recursive iceberg. To put it more precisely: the
development of false belief reasoning competency is itself recursively constructed,
and indeed constructed in a way that tracks the recursive structure of beliefs. At
around the age of four, the child has a remarkable insight: beliefs can be false.
This first-order stage is a fundamental developmental moment, with clear links to
linguistic competency that is well-attested experimentally; the results of false belief
tests are robust under many variations, for example, across countries and various task
manipulations, as shown in the meta-analysis (Wellman et al. 2001) involving 178
individual first-order false belief studies with typically developing children.
Nonetheless, the first-order stage is only the base case in the recursive construction
of beliefs: for while this is the stage where children grasp the crucial idea that beliefs
can be false, they have not yet fully reified the concept of belief. Second-order
competency is the stage when beliefs become just as much a part of the furniture
of the world as anything else, and this reification happens via recursion. Before this
stage, shoes and ships and sealing wax were all part of the world—but now so are
beliefs about shoes and ships and sealing wax, and suddenly the child can have beliefs
about these beliefs, and so on. The world suddenly gets bigger and richer: beliefs
are things, embedded things, Russian-doll-like things, and there are lots of them.
The acquisition of second-order competency is the inductive step in the recursive
construction of belief.
Viewed from this perspective, it makes sense that second-order false belief rea-
soning exhibits logical simplicity and elegance. So to speak, both first-order and
second-order competencies are ‘fixed points’, moments when a significant barrier is
reached and crossed. Both stages are idealizations in the way that useful scientific
concepts often are: they are probably never fully realized in the real world. However,
scientific idealizations are judged at least partly on their utility, and (so far at least)
the concept of false belief competency that develops in two main stages has proved
both useful and robust.
This perspective suggests another way of thinking about JDV reasoning. Recall
that we did not find significant correlations between JDV reasoning and RET. More-
over, the main result of Engberg-Pedersen and Thomsen (2016) was a matched com-
parison between children with ASD and typically developing children, showing that
children with ASD score lower on the JDV test than typically developing children.
This is in line with the well-known fact that the ability to pass first-order false belief
tasks is developed later in children with ASD than in typically developing chil-
dren (see, for example, Baron-Cohen et al. (1985) and many later studies). On the
other hand, such a clear difference is not found in the (admittedly much smaller)
literature on second-order false belief tasks (see Irina Polyanskaya’s Ph.D. thesis
(Polyanskaya 2018) for an overview). This casts further doubt as to whether JDV
reasoning is equivalent to second-order false belief reasoning. Similarly, our result
that second-order false belief does not predict JDV (when controlling for other vari-
ables) gives rise to a similar doubt, and it raises the question of whether passing
the JDV test really involves recursive reasoning, as we have argued is the case with
second-order false belief tests. JDV reasoning certainly does involve taking alternate
perspectives, but it does not seem to require any insight into the recursive structure
of beliefs, or indeed, to depend on having any reified notion of belief at all.
And that’s why JDV is so interesting. Our own work on training children linguistic
recursion suggests that children with ASD do profit from focused linguistic input—
at least in the sense that it leads to improved second-order false belief scores (see
Polyanskaya (2018)). But can such training be generalized to provide help with
the everyday social tasks that children with ASD face? Here, it seems unhelpful to
focus too closely on ToM, and measuring ToM via false belief scores. Second-order
false belief scores are a highly abstract proxy for social competency, and linguistic
recursion is a somewhat marginal phenomenon in spoken language. If the goal is
Another random document with
no related content on Scribd:
The Project Gutenberg eBook of The Corning
Egg Farm book, by Corning himself
This ebook is for the use of anyone anywhere in the United
States and most other parts of the world at no cost and with
almost no restrictions whatsoever. You may copy it, give it away
or re-use it under the terms of the Project Gutenberg License
included with this ebook or online at www.gutenberg.org. If you
are not located in the United States, you will have to check the
laws of the country where you are located before using this
eBook.
Title: The Corning Egg Farm book, by Corning himself
Author: Edward Corning

Gardner Corning
Release date: November 5, 2023 [eBook #72042]
Language: English
Original publication: Bound Brook: The Corning Egg Farm, 1912
Credits: deaurider, Harry Lamé and the Online Distributed

Proofreading Team at https://www.pgdp.net (This file
was produced from images generously made available
by The Internet Archive)
*** START OF THE PROJECT GUTENBERG EBOOK THE

CORNING EGG FARM BOOK, BY CORNING HIMSELF ***
Please see the Transcriber’s Notes
at the end of this text.
Most illustratons may be enlarged
by opening them in a new window or
tab.
CORNING STRAIN UTILITY COCKEREL
Four Months and Twenty Days Old
THE CORNING
E G G FA R M B O O K
BY CORNING HIMSELF
BEING THE COMPLETE AND AUTHENTIC

STORY OF THE CORNING EGG FARM
FROM ITS INCEPTION TO DATE
TOGETHER WITH FULL DESCRIPTION OF THE

METHOD AND SYSTEM THAT HAVE MADE
THIS THE MOST FAMOUS POULTRY
FARM IN THE WORLD
BOUND BROOK, NEW JERSEY

T H E C O R N I N G E G G FA R M
PUBLISHERS
1912
Copyright, 1912, by
GARDNER CORNING
CONTENTS
PAGE
INTRODUCTORY 13
CHAPTER I
The Building of the Corning Egg Farm 21
Started with 60 Buff Rock Eggs 22
More Money in Eggs 25
Adopted White Leghorns 25
First Use of Roosting Closets 27
We Count only Livable Chicks 30
Percentage of Cockerels Low 31
The Great Flock System Succeeds 33
Foreigners Visit the Farm 34
Investigated for Germany 35
Selection of Cockerels 36
Pullets Lay in 129 Days 37
Keeping Down Labor Bill 39
Adopted Hot Water Incubators 40
Why Great Farms Fail 41
CHAPTER II
Egg Farming the Most Profitable Branch of Poultry
Keeping 43
Developing the Great Layer 43
Corning Method in Small Flocks 44
On Large Farms 46
CHAPTER III
What is a Fresh Egg? An Egg Should be Sanitary as
Well as Fresh 48
Manure Drainage to Drink 48
Diseased Meat to Eat 49
As the Food, so the Egg 49
A Perfect Egg a Rarity 50
Unlimited Demand for Quality Eggs 50
CHAPTER IV
Preparation of Eggs for Market 54
CHAPTER V
Selection of the Breed.—The Strain is of Utmost
Importance 58
S. C. White Leghorns Outclass All 59
Line Breeding—Not Inbreeding 61
How Corning Farm Produces Unrelated Cockerels 62
CHAPTER VI
Advantages of Large Flock System—Reduces Cost of
Housing and Economizes in Time and Labor 64
Draughts the Stumbling Block 65
2,000 Birds to a House 66
CHAPTER VII
What is a Winter Layer?—The Properly Hatched and
Reared Pullet 68
Must Feed Green Food 69
CHAPTER VIII
A Great Laying Strain—The Selection of Breeders to
Produce It 71
Eighteen Months Old 71
Trap Nests a Failure 72
Type Reproduces Type 73
CHAPTER IX
Best Time to Hatch 76
Experiment in Late Hatching 78
CHAPTER X
Succulent Green Food—Satisfactory Egg
Production Impossible Without It 80
Sprouted Oats Best 82
How They are Grown on the Farm 82
Timothy and Clover Cut Green 84
CHAPTER XI
Anthracite Coal Ashes—A Substitute for Many More
Expensive Necessities 86
Better Than Charcoal 87
CHAPTER XII
Eggs for Breeding Should be Laid by a Real Yearling
Hen 89
90,000 Orders for 40,000 Eggs 90
CHAPTER XIII
Policing the Farm with Bloodhounds, etc. 92
Shoot First—Investigate Afterward 92
Socrates, the Great Bloodhound 93
CHAPTER XIV
Necessity for Pure Water—An Egg is Chemically 80%
Water 96
Automatic Fountains Essential 96
Hot Water in Cold Weather 97
Hens Drink More in Afternoon 97
CHAPTER XV
Hard Coal Ashes, Oyster Shell, and Grit 99
CHAPTER XVI
Beef Scrap and Green Bone Substitutes for Nature’s
Animal Food 101
Green Cut Bone Nearest Nature 101
CHAPTER XVII
A Time for Everything—Everything on Time 103
Fixed Feeding Hours 103
Four Collections of Eggs Daily 105
Mash Fed in Afternoon 105
CHAPTER XVIII
Incubation on the Corning Egg Farm 106
Hen Reigns Supreme 106
Livable Chicks—Not Numbers 107
Uniform Temperature Most Important 108
Ventilation and Moisture Next 108
Hot Water Machines Best 110
Corning Incubator Cellar Unequaled 111
Eggs Turned from Third to Eighteenth Day 112
103 Degrees Maintained 112
Cool But Never Cold 113
Cover Glass Doors 114
All Good Chicks Hatch in 20 Days 114
Set Incubators Toward Evening 115
Tested Only on Eighteenth Day 116
Moisture 117
Chicks Handled Only Once 117
Baby Chick Business Cruel 118
CHAPTER XIX
Rearing Chicks in Brooder House—The Following
Two Years’ Results Depend Upon Success
in Brooding 121
Corn Not Proper Chick Food 122
Follow Nature’s Teaching 122
A Balanced Food 123
Never Build a Double House 126
Must Drain Chick Runs 127
Concrete Floors Mean Dampness 127
Corning Heated Brooder House 128
Corning Feeds Dry Food Only 129
Three Feeds Daily 129
Green Food Third Day 130
Animal Food Tenth Day 130
Avoid Moving Chicks Often 132
CHAPTER XX
Handling Birds on Range—The Youngsters Must be
Kept Growing All the Time 134
A Corning Wrinkle 135
Grain and Mash Once a Day 137
Plenty of Shade 139
Removed to Laying House Middle of September 140
CHAPTER XXI
Feeding for Eggs—Wholesome Nourishment—Not
Destructive Stimulants 143
Easy Assimilation 143
Perfect Health or No Eggs 144
Abundant Animal Food 144
The Corning Mash the Secret 145
“Egg Foods” Kill Layers 146
Mustard Increases Egg Laying 147
Mustard Increases Fertility 148
4,000 Layers Fed Mustard 149
Mustard Maintains Health 150
Keep Appetite Keen 150
CHAPTER XXII
Breeding Hens During Moult—Coming Breeders Must
be Kept Exercising Through This Period 153
Do Not Overfeed 154
CHAPTER XXIII
Feeding the Breeding Cockerels 156
CHAPTER XXIV
Preparing Surplus Cockerels for Market 157
Must Have Green Food 158
CHAPTER XXV
$6.41 Per Hen Per Year 159
$6.41 Not Extravagant Claim 160
Corning Farm Makes More Than $6.41 161
CHAPTER XXVI
The Buildings on the Corning Egg Farm 163
No. 1, Brooder House, Incubator and Sprouted
Oats Cellars 164
Building No. 2, Work Shop, etc. 167
Building No. 9, Horse Stable 169
Building No. 10, Wagon Shed 170
Building No. 12, Office Building 170
CHAPTER XXVII
Construction of Laying, Breeding, and Breeding
Cockerel Houses 171
Nearly Six Feet from Ground 172
Double Floors 173
Canvas Windows 174
Double Doors 176
Draught-Proof Roosting Closets 177
CHAPTER XXVIII
The Colony Houses—There are Forty-one on the
Farm 180
Cotton Duck Windows 181
CHAPTER XXIX
Materials Required for Laying Houses 182
Bill of Material for the Construction of Colony 183
House
CHAPTER XXX
The Original Thirty Hens 184
CHAPTER XXXI
Egg Records 186
How Corning Farm is Able to Get Great Egg
Records 187
Highest Percentage of Fertility 188
CHAPTER XXXII
Prevention and Treatment of Diseases 190
CHAPTER XXXIII
A Word in Closing 192
Nothing to Hide 193
Illustrations are Photographs 193
The Corning Success 193
Our Advice to Beginners 194
Single Comb White Leghorns Only 194
It’s “Strain” You Want 194
Utility, Not Show Birds 195
Corning Largest Specialty Farm in World 195
Points That Mean Success 196
BUILDINGS ON THE CORNING EGG FARM AND MANY
HANDY DEVICES 198
ILLUSTRATIONS
Corning Strain Utility Cockerel Frontispiece

FACING PAGE
1. Lay-Out of Farm 16
2. Interior Sterile Laying House No. 3, in 1910 22
3. Entrance to Farm in 1909 24
4. As You Approach the Farm, 1911 28
5. Office Building 30
6. Breeding Cockerels, Fall of 1909 34
7. Interior Laying House No. 2, in 1910 38
8. Panoramic View of the Farm 46
9. Thirty Dozen Corning Sanitary Fresh Eggs Ready to
Ship 54
10. The Strain that Makes the Corning Egg Farm
Famous 58
11. Three Sterile Laying Houses Containing 4,500
Pullets 64
12. Interior Laying House No. 1, in 1910 68
13. One of the Breeding Houses just after Mating, 1910 72
14. Sprouted Oats Cellar 78
15. Two-Weeks-Old Chicks in Brooder House Runs 84
16. Yearling Hens in Breeder House before Mating 90
17. “Socrates,” the Great Bloodhound Which Heads the
Corning Kennels 92
18. “Socrates II” and “Diogenes” 94
19. Buster, America’s Greatest Ratter 94
20. Corning Automatic Drinking Fountain 96
21. Part of the Old Incubator Cellar 104
22. Brooder House, Showing Chick Runs 120
23. Old Arrangement of Brooder House 124
24. Chicks Six Weeks Old 128
25. Colony Range Feed and Water Wagon with “Billy” 136
26. Feeding on the Colony Range 140
27. Baskets of Eggs 150
28. Breeding Cockerels, Fall of 1911 156
29. No. 3 Laying House Filled with 1,500 Pullets 158
30. The Workshop on the Corning Egg Farm 162
31. The Celebrated Corning Large-Flock Laying House
No. 3 170
32. Laying House Prepared to Receive 1,500 Pullets
from Range 172
33. One of the Breeding Houses in 1911 174
34. The Corning Colony House 178
35. Breeding House in 1907—The Original Corning
House 182
36. Pullets in Laying House No. 2, Fall of 1911 184
37. Diagrams and Detailed Plans of Buildings, etc. 199
INTRODUCTORY
The Method, and the style of the buildings, evolved and worked
out on The Corning Egg Farm, when put into book form proved so
helpful to so vast a number of poultry keepers, that the sale of this
first literature, which for a time was added to as the months went by,
reached the enormous total of over 140,000 copies in eighteen
months.
The writings were the simple, plain statements of facts, and
enabled others who followed them to reach a success which, until
this System was used, may have been dreamed of, but was never
realized.
The literature from this Farm has gone out over the entire civilized
World, and the visitors, who arrive in ever increasing numbers from
month to month, come from every quarter of the Globe.
The Corning Egg Farm has been written of in periodicals of every
nature, and in almost every language the World over. For the last
twelve months the requests for further, and more explicit, detailed
information relative to breeding and feeding for eggs, the specialty
from which The Corning Egg Farm has never swerved, have become
a demand. So that, after mature deliberation, it was decided to write
the history of The Corning Egg Farm, from its inception to date,
including the work of the last two years, which has never before
been fully published.
“The Corning Egg Farm Book by Corning Himself” is to-day the
only publication giving facts in regard to the Farm and its unique
Method right up to date.
As the book is read it must be borne in mind that, in breeding to
produce a great layer, at first very marked increases in the number of
eggs during the first ten months of laying may be gained. The
general average number of eggs laid each year, from official reports,
is less than 100 per hen. On The Corning Egg Farm, when the
average had reached 143.25 eggs, the next jump, in the following
year, was more than had been expected, and the record of 145.11
eggs for each hen for ten months, though showing an increase
apparently small, in reality was a very great advance indeed.
From this time on, the gain, although representing a narrower
margin of increase, was in reality a much greater achievement. The
trotting horse may serve as an illustration. When Dexter trotted his
famous mile he clipped off a number of seconds from the previous
record, and it seemed as if it would be a matter of considerable time
before his mark would be lowered. But within a comparatively short
time a number of trotters turned off a mile in two-ten, and from this
figure, within a short period, a large company of famous horses had
reached the two-five mark, but every quarter of a second which
reduced this mark meant greater achievement in breeding than was
represented by the reduction of records from two-sixteen to two-five,
and we have not yet seen the horse which, in single harness, without
a running mate, can turn the mile track in two minutes flat.
The Corning Egg Farm realizes that from this on improvement will
be shown by fractional figures, but these fractions will represent a
greater progress than the figures which have gone before.
Two years ago the unequaled results of The Corning Egg Farm
had seemed unsurpassable, but to-day we are able to look back
from higher ground and see the road over which we have traveled to
reach a point very considerably beyond the unequaled position of
two years ago.
It is our hope and aim, year by year, to improve the present
position. The man who believes he has learned all there is to learn is
a failure. The successful man is the one who is sure there is an
opportunity to advance considerably beyond the point he has already
attained, and The Corning Egg Farm believes this to be true, and
has constantly worked with that idea before it.
With an experience back of them of nearly six years the Builders
of The Corning Egg Farm know that this Book furnishes the
necessary guide for success in poultry culture. What has been, and
what is being, done at The Corning Egg Farm is not experimental
work. Successful results follow the Method and System employed as
surely as day follows night. It is no longer necessary for the novice to
try out the various plans proposed to him by the literary poultryman,

in Coherence of Discourse Formal and Conceptual Issues of Language Maxime Amblard Online Ebook Texxtbook Full Chapter PDF

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

in Coherence of Discourse Formal and Conceptual Issues of Language Maxime Amblard Online Ebook Texxtbook Full Chapter PDF

Uploaded by

Copyright:

Available Formats

In coherence of Discourse Formal and

Conceptual Issues of Language Maxime

Language and Law in Professional Discourse Issues and

Historical and Conceptual Issues in Psychology, 3rd

FORMAL LANGUAGE AND AUTOMATA THEORY 2nd Edition Singh

Studies in Formal Linguistics Universal Patterns and

Housing Displacement: Conceptual and Methodological

Endangered Languages and Languages in Danger Issues of

Power and Truth in Political Discourse Language and

Handbook of Formal Analysis and Verification in

More information about this series at http://www.springer.com/series/13376

ISSN 2364-4109 ISSN 2364-4117 (electronic)

• Kata Balogh, Heinrich Heine University Düsseldorf

• Maxime Amblard, Université de Lorraine

Discourse Coherence—From Psychology to Linguistics and Back

Linguistic and Formal Approaches

Maxime Amblard, Michel Musiol, and Manuel Rebuschi

The present volume explores recent advances in modeling discourse processes,

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 1

Developing systematic analysis methodologies for evaluating verbal communica-

(3) theory of mind/mental state attribution that encompasses processes of under-

2 Approaches to Language and the Linguistic Approach

Cognitive Psychotherapy treats verbal material more abstractly and independently

Coherence and incoherence are normative concepts. Identifying inconsistencies in

Studies on discourse incoherence commonly use corpora involving patients diag-

4 Presentation of the Volume

differences in reasoning supported by non-logical inferences leading to misunder-

count as an assertion or expression of beliefs may, in another, signify an expression

Cutting, J. (1979). Memory in functional psychosis. Journal of Neurology, Neurosurgery and

Patrick Blackburn, Torben Braüner, and Irina Polyanskaya

Abstract In a study involving 62 Danish children with autism spectrum disorder,

Keywords Recursion · Autism spectrum disorder · Sentential complements ·

2 Second-Order False Belief Tasks

According to the fifth edition of the American Psychiatric Association’s Diagnostic

interface, see Szabó (2013).

Tager-Flusberg and Joseph only discuss first-order reasoning. We hypothesized that

One final remark: as logics capable of handling recursive embeddings of beliefs

In this section, we describe the correlation component of our study (Polyanskaya

reading/writing skills required to take it. In addition, we asked teachers to complete

Fig. 1 A RET example

5 RET and Second-Order False Belief Reasoning

As we shall now see, RET is a significant predictor of success in second-order false

Table 1 Kendall’s τb correlations (Sig. Levels: * p <0.05, ** p <0.01)

Table 2 Regression analysis

We are now ready to consider a somewhat puzzling result. As we have already

7 The JDV Test

Fig. 2 JDV score histogram

Title: The Corning Egg Farm book, by Corning himself

Author: Edward Corning

Release date: November 5, 2023 [eBook #72042]

Original publication: Bound Brook: The Corning Egg Farm, 1912

Credits: deaurider, Harry Lamé and the Online Distributed

*** START OF THE PROJECT GUTENBERG EBOOK THE

BEING THE COMPLETE AND AUTHENTIC

TOGETHER WITH FULL DESCRIPTION OF THE

BOUND BROOK, NEW JERSEY

Corning Strain Utility Cockerel Frontispiece

You might also like