WOLFGANG WILDGEN

NATURAL ONTOLOGIES AND SEMANTIC ROLES IN SENTENCES

1. STARTING FROM HARTMANN ’ S ONTOLOGY

My central concern in the treatment of Hartmann’s “New Ontology” will

be the role of language and other “symbolic forms” (in the terms of Cas-
sirer’s “Philosophie der symbolischen Formen”, 1923–1929). The major
strata (Schichten) in Hartmann’s ontology are called “well-know in their
contours” by himself.
Nun bildet die Mannigfaltigkeit der Formen offenbar ein Stufenreich, dessen Rangordnung
im groben wohlbekannt ist: Ding, Pflanze, Tier, Mensch, Gemeinschaft. (Hartmann, 1964:
36)1

Although the transitions to plants, to animals and mainly the more recent
one to man are highly problematic, there is some consensus that they are
relevant as transitions. The transition between man and the community
of men or even to historically evolving “Volksgemeinschaften” (ibid.) is
highly controversial on one side because the time scale is evolutionarily
so narrow: from millions of years almost to centuries, on the other side
because this transition fails to be inclusive: If man shares features (e.g.,
physiological and genetic features) with animals and plants this does not
hold for the couple: man – community. In 1933 Hartmann tells us in his
“Systematische Selbstdarstellung”.
Der Gemeingeist ist von keinem Gemeinbewußtsein getragen, sondern nur von individu-
ellem Bewußtsein. Und dieses gerade hat, so zeigte sich, keineswegs die Tragkraft für ihn.
So bleibt seine Seinsweise in aller Durchleuchtung des Phänomens eine metaphysisch-
rätselvolle. (Hartmann, 1933/1955: 35)2

It is, nevertheless, this transition which separates “Seele” (soul) and

“Geist”(mind) and circumscribes the area of the “Geisteswissenschaften”
to which philosophy and semiotics belong. “Geist” may be characterized
as transcendental in relation to the (individual) “soul”.
Er [Geist] transzendiert die Enge der Bewußtseinssphäre in der Mitteilbarkeit, im Über-
greifen von Subjekt zu Subjekt, in der Tradierbarkeit von Generation zu Generation;

Axiomathes 12: 171–193, 2001.

© 2002 Kluwer Academic Publishers. Printed in the Netherlands.
172 WOLFGANG WILDGEN

so wie andererseits auch der geschlossene Kreis des Einzelbewußtseins seinen inhalt-
lichen Reichtum nicht aus sich allein schöpft, sondern aus dem geistigen Gemeingut der
Lebenssphäre, in die es hineinwächst und an die es sich angleicht. (Hartmann, 1940/1955:
74)3

The phenomena pertaining to the stratum of “Geist” (mind) correspond

roughly to those called “symbolic forms” by Cassirer. Hartmann says:
Dieses Seinsgebiet ist das der Sprache und des Rechts, der Sitte und der politischen
Bewegung, des Wissens und der Künste. (ibid.)4

It would be an interesting endeavor to compare Hartmann’s comments

on the transition between “Seele” and “Geist” with Cassirer’s elaborated
philosophy of symbolic forms as an objectivation of the individual think-
ing, feeling and experience. Such a comparison would have to consider
at least the other thinkers between 1900 and 1930 contributing to the
same question: mainly Husserl, Heidegger and Carnap; it is, therefore, not
possible in the context of this paper.5
Both Hartmann and Cassirer take the philosophy of Kant as the point
of departure of modern philosophy and both are critical towards the “Neu-
kantianismus” of the Marburg-school. Cassirer turns systematically to
other sources of knowledge than science in his philosophy of symbolic
forms, where he considers language (first volume), myth (second volume)
and roughly cognition (third volume). Hartmann mentions three major
sources in the establishment of a “new ontology”, which has to start from
basic human experiences:
Es setzt die ganze Breite der Erfahrung voraus, sowohl der des Alltags und des prakt-
ischen Lebens, als auch der wissenschaftlichen. Ja, man kann hinzufügen, es setzt auch
die philosophische Erfahrung voraus, diejenige nämlich, welche in dem geschichtlichen
Gange menschlicher Denkarbeit als eine lange Reihe von Versuchen, Fehlschlägen und
Selbstkorrekturen verzeichnet ist. (Hartmann, 1964: 18)6

To this broad experiential basis is added the critical reflexion (in Kant’s
sense). This basis corresponds roughly to the one of Cassirer. Cassirer is
even more engaged in assembling of current scientific results and in the
history of scientific thought (cf. his four volumes: “Das Erkenntnisprob-
lem in der Philosophie und Wissenschaft der Neueren Zeit”). The major
difference concerns the role of mathematics in the sciences themselves
(at least since Galilei) and in philosophy (at least since Descartes). In
the search for possible general laws and basic principles, mathematics
had taken over the role of medieval logics. Although Cassirer is not eu-
phoric when he discusses the application of mathematical thought to the
“Geisteswissenschaft” (cf. his article on group theory and the psychology
of vision, Cassirer, 1944), he is aware that any quest of universals, basic
 NATURAL ONTOLOGIES AND SEMANTIC ROLES IN SENTENCES 173

TABLE I
Distance-curve between ontological and epistemological categories.

principles will have to use (or properly develop) mathematical tools. Hart-
mann seems to restrict the help of mathematics to the domain of quantity
and in its application to physics. In his article “Die Erkenntnis im Lichte
der Ontologie” (Hartmann, 1955: 149) he shows a diagram where a curve
depicts the distance between epistemological categories and ontological
categories. Table I shows the graphic.
The distance is small for “Geist” and for “Anorganisches”, but the
reason is a different one. In the domain of “Geisteswissenschaften”, lan-
guage and every day knowledge brings together “being” and “knowledge”;
in the domain of physics (quantitative) mathematics does the job. The
fact that since the program of Erlangen (Klein’s theory of invariants) and
since the rise of topology and qualitative calculus (Poincaré) a new type of
qualitative mathematics had been developed, is not acknowledged in Hart-
mann’s philosophy of science. Cassirer, Carnap, Lewin and many others
reacted to this evolution in the second half of the 19th and the beginning
of the 20th century (cf. Wildgen, 2001). The general systems theory of
Bertalanffy, the cybernetics of Wiener, Prigogine’s theory of dissipative
systems since the 1940s have proposed unifying, mathematically rooted
systems for the problem of multi-stratum categories, which lies at the
174 WOLFGANG WILDGEN

heart of Hartmann’s “New Ontology”. The catastrophe theoretic and chaos

theoretic models developed in the 70s and 80s followed a similar path.
Hartmann’s fundamental reflections on ontological strata and transtrata
categories and their growing complexity were contemporary to these new
developments; but his reflections had no significant effect on them. The in-
novative impetus of Hartmann’s new ontology, which asked to be rooted in
current scientific research, was lost insofar as the mathematically organised
disciplines originated their own “ontology” and own “categories”.7
I shall not develop this historical criticism of Hartmann, but propose an
ontology rooted in catastrophe theory and in a type of general semantics
adapted to the formal schemata put forward in catastrophe theory. In a
first step I will sketch some ontological presupposition implicit in such a
treatment and relate it to Hartmann’s critical “new ontology”. The central
concern of dynamical models of the world, the mind, human languages,
is the processual aspect. In this we are still in accordance with Hartmann,
who says (pointing to pre-socratic philosophers):
Das Werden ist vielmehr die allgemeine Form des Realen. (Hartmann, 1955: 99)8

It cannot be reduced to states and transitions between states:

Und eben darum geht das Wesen des Prozesses auch wiederum nicht in der Ausein-
andergezogenheit der Stadien auf. Die Stadien sind nur flüchtige Durchgangsstadien,
Augenblickzustände; aber indem die Zustände kommen und gehen, beharrt der Prozeß.
(Hartmann, 1955: 100) 9

In the domain of cognition (i.e., on the stratum “Seele” in Hartmann’s

hierarchy) the process character is also basic:
Als Bewußtseinsform ist der Prozeß eine eminente Anschauungs- und Erlebnisform,
keineswegs erst eine Form des eindringenden Begreifens. (ibid: 101)10

My strategy in the following will be to consider first the conditions for

the identity and the categorization of processes in general. To this con-
cern the results of catastrophe theory will be used. Second, processes can
have different substrata, forces, etc., which materialize them. Any such
interpretation of the mathematical form (f.i., a type of catastrophe, or a
schema derived from it) implies an ontology. If we move to language, the
concept of sign or representation implies an ontological heterogeneity, i.e.,
two, three or more strata, elements of which are correlated in the sign (i.e.,
the very concept of representation as such implies a stratification). If we
consider a triadic sign structure, then three strata have to be distinguished:
– the stratum of the referent (in the world),
– the stratum of the interpretant (individual or social consciousness),
– the stratum of the representamen (the body of the sign itself).
 NATURAL ONTOLOGIES AND SEMANTIC ROLES IN SENTENCES 175

TABLE II
Triadic extraction of stability (unity) from processes.

In the case of linguistic signs basic realizations are acoustic processes (cor-
related with organo-genetic, auditory and mental events), writing, gestures
and many technical transformations of them.
As the semiotic discussion since antiquity shows, the ontological strata
underlying sign-usage are highly controversial. If we adopt a triadic struc-
ture, we choose the most plausible one, but a definition of the sign cannot
be used as an a priori statement from which an ontology may be derived.
We can only accumulate evolutionary, cognitive, sociological arguments in
favor of such a stratification.11 The basic scenario is given in Table II. The
underlying invariant is the notion of process (cf. above).
Ontologically the three components in Table II overlap, as the identi-
fication of phonemes or words in a sound flow is basically of type A. It
is the specific tuning of the recognition process to one language which
constitutes a phenomenal type B.
The results of language learning (type C) shape the cognitive structure
of individual perception and memory (type B) and the stable recognition
of things and event-types (type A) allows for a system of meanings and the
correlation of meanings with sign-events (type C). The complexity of these
interactions was acknowledged by Hartmann insofar as the “Idealsphäre”
(ideal sphere) on one side and the “Realsphäre” (real sphere) on the other
side intersect. Specifically in the area of intersection related to “Seele”
(soul) and “Geist” (mind) the “logische Sphäre” (logical sphere) L and the
“Erkenntnissphäre” (epistemological sphere) E cover one another. In the
zone of maximal intersection all four domains interact as Figure 1 shows
(cf. Lichter, 1964: 58).
176 WOLFGANG WILDGEN

Figure 1. Modes of intersection in Hartmann’s “Kategorialanalyse”.

2. THE ROLE OF ONTOLOGIES IN A PHILOSOPHY OF THE MIND AND

LANGUAGE

Hartmann arrives to the specific perspective he takes by reintegrating the

subject (of cognition, judgement, knowledge) into the world as a proper
part of it. Nevertheless, the problems of epistemology raised by Descartes’
skepticism and Kant’s “critique of pure reason” are still present. In his
“New Ontology” Hartmann considers a duplication of the ontology, which
contains the human mind and its stratification in the sphere of the mind
(“Geist”). The duplicates which are now “objects” of cognition, know-
ledge, are called contents (“Inhalt”). They focus on new structures based
on correlations (“Zuordnung”), a scale of knowledge (“Erkenntnisstufen”),
beginning with simple perception and ending with comprehension (“Be-
greifen”), science and critical self-control (cf. Hartmann, 1964: 109 f.).
Qualities belong to the basic epistemological stratum (ibid.), whereas
categories (in the sense of Kant’s table of categories) are bound to the
generality only reached in the judgement. This duplication triggers the
problem of correspondence between real categories (“Realkategorien”)
and content categories (“Inhaltskategorien”). There are basic differences:
the infinity and continuity of real space and time is reduced to finite-
ness and discreteness in the content categories. Perspective, selectivity and
vagueness are general types of deformations characteristic for this cor-
respondence relation. His new ontology opens the way for a differential
analysis of existing categorizations, which had been a forbidden terrain in
traditional treatments of the problem (ibid: 113).
The differences between levels of categorization become even more
dramatic, if one considers the linguistic nature of human judgements
and of the corresponding knowledge structures and the diversity of lan-
 NATURAL ONTOLOGIES AND SEMANTIC ROLES IN SENTENCES 177

guages in the domain of lexical categorization. The apparently simple

mapping between real categories (in Hartmann’s sense), content categor-
ies, linguistic categories, and specific categorizations in the lexicons and
grammars of natural languages is lost, if we leave the abstract level of a
quasi-universal epistemological sphere. Hartmann did not dare to enter
this area of complication, and, in this respect, he is a true follower of
Kant. Herder had already challenged Kant’s language-free treatment of
epistemology in his “Metakritik” and Cassirer took up this controversy in
his “Philosophie der symbolischen Formen”.
Before I propose an integrative view of ontology in Section 3, I will just
mention four developments which were prior, contemporary, and posterior
to Hartmann’s new ontology.
(a) In his “Der logische Aufbau der Welt” (1928) Rudolf Carnap
proposed a hierarchically organized “Konstitutionssystem” inspired by
Husserl’s phenomenology. For him the act of cognition and comprehension
is the starting point from which an ontological stratification may be derived
(he would not call it “ontological”, however):

– objects in the own mind (the phenomenological starting point),

– physical objects (contents of the mind with reference to the world),
– objects belonging to other minds,
– abstract objects (e.g., culture, society, religion).
In Hartmann’s perspective the content categories are the starting line, then
the different spheres (matter/life/soul/mind = “Geist”) are assessed. Al-
though Carnap’s position seems to be opposed to Hartmann’s it could be a
variant under the perspective of the subject of cognition.
(b) Under the influence of gestalt-psychology (e.g., the concepts of
valence and field; cf. Wildgen, 2001), Gibson and the school of “ecolo-
gical psychology” (cf. Kelso, 1997) developed another perspective, where
the psychophysical interface is the turning point from which external real-
ity/other minds and internal reality may be reached. Such a stratification
was presupposed in a “realistic” model of meaning in Wildgen (1994) and
I will come back to it in Section 3.
(c) In Artificial Intelligence (AI) and Natural Language Processing
(NLP), which exploited the possibilities of the digital computer as a system
of knowledge representation, another strategy has been followed. First,
rather ad hoc ontologies abstracted from one language (mainly English)
were devised to build up an architecture of contents.12 As more knowledge
domains and different languages were considered, a theoretical discussion
was called forth in search for general/optimal ontologies. Thus, Bateman
(1992) distinguishes three types of ontologies:
178 WOLFGANG WILDGEN

– (non-linguistic) conceptual ontologies based on common sense

(whatever this may be). Empirically this ontology could be elaborated
in psychology or sociology.
– Linguistic ontologies, based on the lexico-grammars of natural lan-
guages.
– Interface ontologies which consider linguistic information to be just a
subset of conceptual information.
(d) In Hartmann’s ontology, the “Realsphäre” has a stratum called
“Geist” and is separated from “Seele”. In the realm of “cognitive se-
mantics” (since the 1980s), it seemed as if every thing was reduced to
the brain (i.e., to individual cognition) and that further categorizations
could only subdivide the mental. A typical example is the model of
“mental maps” by Fauconnier (1984) and of “Blending” by Fauconnier
and Turner (2002). Fauconnier distinguishes different domains to which
“mental spaces” belong and reconstructs them on the basis of informa-
tion contained in linguistic expressions. Mental spaces are connected to
discourse (in reality) through pragmatic connectors (cf. Fauconnier, 1984:
32). In “blending”, different levels of mental spaces interact in a lattice-
like structure. Two mental spaces from different domains (e.g., language
and picture) are linked by a generalized mental space which establishes
their comparability; this space is called “generic space”. If two mental
spaces are blended, the result (the blend) inherits some of the features of
both (enhanced by the generic space) and may create new features by the
emergence of blending-effects.
The basic configuration of four spaces and five mapping relations is
shown in Figure 2 (cf. Fauconnier, 1999).
The spaces belong ontologically to the mental (“Seele” in Hartmann’s
terminology), but methodologically insofar as they are inferred from
linguistic usage to the symbolic-collective (“Geist” in Hartmann’s termin-
ology). The generic space may in general cases come near to logical or
epistemological categories and thus intersect with the “Idealsphäre” in
Hartmann, whereas the blends, if they are innovative, create spaces beyond
the “real”. But what reality do all these spaces have? From a linguistic
point of view, Fauconnier argues that every reference pertains to men-
tal entities, which may be separated as belonging to different domains
between which mappings and metaphors operate. This may be true at the
level of discourse, where the speaker/hearer and their situation are the only
anchoring domains which count. In the semantics of words and sentences
a different, more basic, ontological classification can be inferred, which
recovers the intuitions of philosophical ontology in the sense of Hartmann.
Moreover, and this corresponds to Hartmann’s basic goal, our scientific
 NATURAL ONTOLOGIES AND SEMANTIC ROLES IN SENTENCES 179

Figure 2. Basic blending-structure.

endeavor has an asymptotic tendency to categorize different strata of the

real world adequately (even if this is prima facie less important in everyday
discourse). Therefore, the possibility of a proper ontological grounding of
scientific language is rather a regulatory principle than a linguistic reality.
Together these proposals show that the empirical and intuitive evidence
which makes a specific ontology plausible has different sources:
– common sense or every-day realism,
– psychological and sociological explanations of common sense,
– languages as basic inventories of ontological assumptions and categor-
izations,
– results of the natural sciences which go systematically beyond com-
mon sense realism,
– coherence criteria of any basic system of assumptions and formal
architectures of such systems (ideally in a mathematical format).
In the following section I shall propose a set of principles for a stratified
ontology which uses catastrophe theory as formal architecture, and which
integrates results of physical dynamics, cognitive schematizations and ba-
180 WOLFGANG WILDGEN

sic sentence patterns. The fact that catastrophe theory is a generalization

over basic insights into the stability and dynamics of systems in evolution
(mainly in the domain of natural sciences) is not further elaborated (cf.
Wildgen, 1982; Petitot, 1992; Wildgen, 1994, 1999). The ontology I pro-
pose has been developed in the context of research in cognitive semantics
(cf. Wildgen, 1994: chapter 5) and is presented here in a slightly modified
version.

3. PROPOSAL FOR A SEMANTIC STRATIFICATION BASED ON

CATASTROPHE THEORY

The strata I will propose are ontological only in a derived sense. The se-
mantics of natural languages imply an ontological hierarchy by the types
of verbs and the semantic roles (valence patterns) these select. As we start
from verbs and basic sentences which have these verbs at their center, the
stratification is not a static but a dynamic one. It concerns “gestalts”, rela-
tions, which secondarily ask for fillers, arguments, valence governed noun
phrases or in a more general fashion for semantic roles and deep cases (cf.
Wildgen, 1985: chapter 2 for a description of basic issues in the semantics
of “deep” cases). By the choice of this procedure in the reconstruction of
an ontological stratification we adopt the perspective of ecological realism
and not the constructivist perspective of Carnap. The procedure seems to
be compatible with Hartmann’s intuitions of realism, but it goes beyond
the stratification he has proposed. Therefore the systematic development
shown in the following sections is not meant as a reconstruction of Hart-
mann’s “Kategorialanalyse”; it can easily be associated with the program
of morphodynamics proposed by René Thom and elaborated in Wildgen
(1982) and later in Petitot-Cocorda (1992) and Wildgen (1994). A first
semantic principle stratifies the domains of interpretation. This principle
is the basis of the semantic characterization of verbs in Wildgen (1994:
Chapters 3 and 9).

Semantic principle of stratified domains of interpretation

Four major stratified domains can be distinguished: 1. Locomotion in
space. 2. External action/interaction. 3. Internal action. 4. Change in a
quality space which are further subdivided in Tables III–VI.

Locomotion in space depends basically on physical laws (gravitation,

dynamics of simple and double pendulums) and thus refers to Hartmann’s
category “matter” and inorganic laws (cf. Table I and Figure 1). The psy-
chophysics of bodily locomotion involve a biological system, the body,
 NATURAL ONTOLOGIES AND SEMANTIC ROLES IN SENTENCES 181

TABLE III
Subdivisions of the first domain.

which reduces the degrees of freedom of double-pendulums, introduces

barriers and filters and thus selects specific motor-program (stratum: life),
which are controlled by brain-centers (stratum: soul). Moreover, the modes
of locomotion are learned (this involves socialization) and have individual
features, which allow the identification of individuals (these features pre-
suppose the stratum called “Geist”). Domain (1) covers and separates all
strata of Hartmann, but the most prominent in terms of dynamics is the
stratum of physical laws of motion.
The domain (2) of external action/interaction is complex insofar the
manipulated object is typically governed by physical laws (gravitation,
solidity), the action is, however, dominated by intentional control, purpose,
needs, functions, etc. The center of this domain is, therefore, the stratum
of life, where these biological functions are grounded.
The domain (3) of internal action has the objects on the strata: mat-
ter and life, as contents (as Hartmann pointed out correctly). Its basic
stratum is the brain (and other cognitive subsystems) and it corresponds
to the stratum: soul (“Seele”) in Hartmann’s ontology. The domain (4) of
qualia finally has the soul, the organs of categorical perception, memory,
imagination, and creative invention as substratum, but it depends heav-
ily on learning, socialization, and language. Its center is, therefore, the
stratum “Geist” and all mappings (e.g., brain topologies) have a strong
conventional character. Qualia are, as Peirce pointed out, basically signs.
Under this interpretation the stratification of domains which grounds the
semantics of human language, is compatible with Hartmann’s proposals.
He just did not go far enough into the cognitive and linguistic realms and
was fixed to an epistemology critically dependent on the natural science
(i.e., on the kind of “Naturphilosophie” they contain implicitly).
The same type of process may have a richer dynamical context and
lead to action and interaction. As the perspective changes from one agent,
to configurations of agents and objects/other agents, one may assume a
further domain (or continuation of the first one with higher complexity).
182 WOLFGANG WILDGEN

TABLE IV
Subdivisions of the second domain.

TABLE V
Subdivisions of the third domain.

Action and interaction stand ontologically between physical loco-

motion (1), which governs parts of them, and internal (intentional) pro-
cesses (3) which direct the action. The effect is often a change of quality
(4). These processes are typically mixed, i.e., the different roles in an
action/interaction scenario operate on different strata.
In domain 3 the processes are strictly internal within a body or a
cognitive system; we cannot observe them directly in other people and can-
not refer to them objectively. These processes have, however, perceivable
traces (in the behavior of the individual) and we can linguistically label
such a process and tell the event to our audience. The processes of domain
3 also have another peculiar property. They are the basis of a modality
scale. The subdivision into the domains 1, 2, 3 hides the transitions which
are rather smooth. Thus 2.3 is very close to 3.1 since the communicative
act contains the aspect of perceiving a sign and of emitting one; so it is
perceptual (3.1) and external (2.3) at the same time.
The change can be mono-valent (the quality appears), bivalent (it
changes to the other pole of the scale), or trivalent (with an intermedi-
ate phase on the quality scale). In Wildgen (1994: chapter 3) schemata
for locomotion, action and interaction were constructed as configurations
which underlie the semantics of verbs and their valence. If we apply prin-
ciple 1, these schemata can be interpreted in a different way, depending on
the stratum, i.e., in a body-centered space-time (domain 1), in a space of
 NATURAL ONTOLOGIES AND SEMANTIC ROLES IN SENTENCES 183

TABLE VI
Subdivisions of the fourth domain.

action/interaction (domain 2), in a mental space (domain 3) or in a quality

space (domain 4). The mathematical parameters contained in the schemat-
ization are interpreted by the dimensions of these spaces in the same way as
mathematical models are interpreted differently for specific “real” systems
in physics or chemistry. In this sense the domains introduced in principle
1 flesh out the formal schemata, they define the phenomenological found-
ation of the schemata and ensure that they are more than a mathematical
construct on the one hand, or as an abstract level of a purely linguistic
categorization on the other.
For every domain we may distinguish maximum schemata and partial
schemata. The bordering or abutment relation between elementary cata-
strophes, whereby catastrophes lower down the hierarchy appear locally
within higher ones, provides the formal basis for a general principle which
relates the maximum schema to a partial one (see Wildgen, 1982: 93–97).
This principle is called the “principle of dynamic inclusion” and will be
applied in the following section.

Semantic principle of dynamic inclusion

A maximum schema (in a specific domain, cf. principle 1) contains topo-
logically partial schemata which are simpler (have a smaller number of
attractors) and are constituents of the maximum schema.

As a corollary we can state that if a non-maximum schema appears

in a certain domain further (partial) schemata may be inferred and used
in the organization and interpretation of larger utterances and texts. Only
the maximum schemata will be enumerated in order to illustrate the
dependence between domains and schemata.
184 WOLFGANG WILDGEN

4. A BASIC SET OF SEMANTIC ROLES

As a consequence of the first semantic principle we get different types of

simple process-scenarios in every domain. These process-scenarios define
a set of semantic roles for every domain. Within my framework, semantic
roles are not purely linguistic labels, since they can be interpreted in terms
of a cognitive theory of language. Thus their relation to surface phenomena
like prepositions, morphological cases and word-order is less immediate
than in a case-grammar of the type proposed by Fillmore (1968) and
further elaborated in Fillmore (1977). Formally, my semantic roles are
considered as variables which receive different interpretations in every
domain. The invariant is not a syntactic or a semantic feature but the
topological configuration of which a semantic role is a partial domain.
Although I use traditional labels, the content of these labels is not given by
everyday use or by the usage in recent traditions. I systematically depart
from case-theory insofar as my primary criterion is that of dynamic config-
uration. The possible dynamic configurations are nested and hierarchically
structured (cf. principle 2). I distinguish:
– primary agents (they are the foundation of the process and do not
disappear in the process): A, P;
– secondary agents (they appear and disappear in the process): I, B.
In my terminology Source-Goal are “allo-cases” since they interpret the
configuration of two opposed poles in the domain of spatial locomotion.
The “casemes” defined by the configurational criterion are called:
– A (Agent) – P (Patient) (primary roles)
– I (Intermediary) – B (Binding force) (secondary roles).
The label I summarizes a plurality of forces which are linearly intermediate
between A and P. Depending on the domain of interpretation, I can be a
path (interlocal locomotion), a metastable phase on a quality scale (quality
space), an instrument (action space) or an object (change of possession).
The role B (binding force) has a rather variable realization. Configuration-
ally it is an intermediary force parallel to the primary sequence A-I-P. It
therefore calls for a second dimension in state space (cf. Wildgen, 1982:
85–92). It can be parallel to A (a helper of the agent), to P (a beneficiary
of the event) and to I (a secondary instrument, a medium of exchange).13

4.1. The domain: locomotion in space-time

Foreground and background must be distinguished (Langacker, 1987, calls
these traditional terms ‘trajectory’ and ‘landmark’). The background (land-
mark) is typically a domain in space-time; syntactically it is a prepositional
 NATURAL ONTOLOGIES AND SEMANTIC ROLES IN SENTENCES 185

Figure 3. The maximal schema of locomotion and partial schemata.

Figure 4. The maximal schemata of action and interaction.

or adverbial phrase and does not contribute to the thematic grid (the
configuration of semantic roles).
Locomotion may be simple (linear) or include the transition through a
frontier or even several, linearly arranged frontiers (on a path). The max-
imum configuration is one with the three roles: A (agent), P (patient), I
(intermediary force). A possible elaboration contains one or more domains
on the path through which the intermediary force goes when it comes from
the source and before it reaches the goal. The maximum configuration has
three roles, as Figure 3 shows. Partial configurations have two or just one
role (attractor).

4.2. The domain: action and interaction

The maximum configuration is the schema of transfer (or of instrumental
action which is the symmetric variant of it). Figure 4 shows the two
schemata.
The two variants have the same thematic grid (A-I-P). If principle 2
is applied implicit relations between the two scenarios and their thematic
grid can be established.
In Figure 5 the elaborated schema which refers to a two-dimensional
behavior space is schematically represented.
186 WOLFGANG WILDGEN

TABLE VII
The maximal schema and the levels of reduction.

Figure 5. The elaborated schema (two-dimensional behavior space).

The fourth participant is called the binding force and it can be inter-
preted as a helper (i.e., a secondary agent in the tradition of narratology)
or a beneficiary (a secondary patient). The four-valent scenario can be fully
realized in the scenario of instrumental sending:

Example:
(i) Albert (A: source) sends Indra (I: secondary agent) with British
Airways (B: helper) to Paris (P: goal)
The intermediary force can also be an object exchanged, or a primary
instrument.

Examples:
(ii) Andrea (A) sends a letter (I) to her friend (P) by airmail (B).
(iii) Annabel (A) gives an interview (I) to the press (P) by telephone (B).
(iv) Anne (A) propels the arrow (I) towards the tree (P) with a bow (B).
The further elaboration is characterized by a completion of the basic
schema to a symmetrical configuration. In this completion the two inter-
mediate roles may be split so that the values of the intermediate roles
become different. This process is typical of possessive interaction (see
Table II above) where the object bought/sold and its equivalent, the money,
 NATURAL ONTOLOGIES AND SEMANTIC ROLES IN SENTENCES 187

Figure 6. The basic schema of communicative action.

fill two symmetrical, although different roles in the schema. The basic
configuration is that of giving, completed towards a mutual gift.
As in the previous case an intermediate secondary force (e.g., a trans-
mitter) may be introduced. In many realizations, e.g. simple sentences,
partial schemata are preferred, i.e., some other part of the schema is left
unrealized. Thus in the following sentences we find a reduced realization
of the basic schema:

(v) My mother bought me a book.

(vi) Charles bought a bicycle from his friend for $50.

In (v) the source (A) is not mentioned but the buyer (the binding force
B: my mother), the object (I) and the Patient (P: me) are realized; in (vi)
the Source (A: from his friend), the object (I1 ), the price (the equivalent
to the object) I2 and the beneficiary, who is identical with the buyer (“for
himself”, B = P), are mentioned.

4.3. The domain: communicative and perceptual action

The configuration is similar to those already discussed. We can distinguish
between emissive actions, where perceivable events are produced and re-
ceptive actions, where such events are received. If both partial schemata
combine we have a transfer of perceivable units, signal transmission; if
this transmission is mutual and reciprocal we have sign communication.
If a symbolic instrument, a system of conventional signs, is put to work,
a binding force is added. Language as a system is such a symbolic instru-
ment. The roles which are defined by their place in the configuration have
a somewhat different content. Figure 6 shows the basic configuration.
If we take a closer look at the dynamics of the event, we notice import-
ant differences between communicative action (domain 2.3) and the basic
domains 2.2 (change of possession) and 1 (physical action):
188 WOLFGANG WILDGEN

Figure 7. The basic schema. of mental action.

– The message is not lost by the sender if he emits it. Rather, he sends a
duplicate; in a similar vein the receiver creates an analogous message,
using the information he receives and mental knowledge.
– The intermediary role B (the binding force, the code) is a neces-
sary constituent for the transfer which could not take place without
it. Furthermore this force is very rich and complicated. Whereas the
Agent and the Patient are individuals, the language system has a social,
supra-individual and, therefore, abstract nature.
In perception, the object received can be either a sign or simply a per-
cept (some natural input to the sensory organs). The sensory inputs which
are continually entering our sensory organs are the background of sign-
reception. At an intermediate level our attention is focused on a specific
percept; we see, hear, smell something specific. The topological scenario
is that of capture.
These basic deviations from external processes become even more
prominent if we analyze what is going on in mental action.

4.4. The domain: mental action

The new phenomena at this stratum are:
– the semantic closure of the mind on itself, this strange feature was
emphasized by Maturana (1980) and other theoreticians of the brain,
– the self-referential nature of mental processes,
– the overwhelming importance of cognitive contexts, i.e., of memory,
knowledge, emotional disposition, attitudes, personality traits, etc.
These basic characteristics are schematically represented in a specific form
of the maximum schema, which is given in Figure 7.
The different phases of the process are labeled:
– egressive (emission): the mind produces an idea, an emotion, an
attitude,
 NATURAL ONTOLOGIES AND SEMANTIC ROLES IN SENTENCES 189

– ingressive (reception): the mind receives, retains, stabilizes an idea, an

emotion, an attitude,
– self-referentiality: the mind produces and receives (from itself) an
idea, an emotion, an attitude.
The egressive and ingressive schemata are only partial pictures of self-
referentiality which is more complete. The actual processes can become
part of the permanent structure of the mind and parts of the permanent
structure can in turn be actualized. This mode is called resultative. The
pure form of the resultative phase is the stock of persistent ideas, emotions,
etc. in the mind.

4.5. The domain: change in a quality space

The configurations are similar to those in Section 4.2, the difference being
that partial schemata are more frequent and that the elaboration with a third
(intermediate) quality are very rare. Parallel to the traditional distinction in
mathematical linguistics (cf. Marcus, 1967: 4–20), we can introduce two
pairs; A vs. non-A (privation of A) and A vs. CA (bimodal proportional
opposition of A to its complement CA). In the first case we consider only
a partial scenario while the complementary state is left undetermined. In a
proportional opposition both qualitative states are present; the change from
one quality to the other is moved into the foreground.
1. Privation
stop to be A (become non-A),
begin to be A (stop to be non-A);
2. Proportional
stop to be A (leave the domain A and become CA, i.e., change from A
to CA),
begin to be A (leave the domain CA and become A).
We can easily see that the first schema is a proper part of the second one.
Thus the relation between the two falls under principle 2.

5. NATURAL TOPOLOGY OF THE HUMAN MIND : SUMMARY

Humans and their minds are the center of a natural topology which organ-
izes the domains introduced before. Table VIII shows the ordinal scale of
domains, relative to the mind of Ego as it has been developed in the last
sections.
An event/change in the ecology of the speaker can be perceived as
belonging to a specific domain, for instance to the basic domain of loco-
motion. In the utterance it can, however, be embedded in another domain,
190 WOLFGANG WILDGEN

Figure 8. Basic processes in a quality space.

TABLE VIII
The scale of domains.

such as perception, imagination, remembering or it can be considered on

behalf of some qualitative result etc. This means that the speaker (and with
him the listener) can move on the scale shown in Table VIII. The projec-
tion of “real” events onto such a scale will be called a “modality scale”.
Despite the introduction of this freedom from “real” events introduced by
movements on the modality scale, the semantics are still “realistic” insofar
as the transformations of events belonging to the ecology of the speaker
are not independent from the bodily basis of meaning but are rooted in it
and inherit their stability from these roots. We predict that in those cases
where the “realistic” links are lost, the coordination between speaker and
listener, or even the stability of “inner” language in the speaker himself, is
also lost and language degenerates (it becomes noise). If we have a single
event (at a center level on the modality scale), e.g., a process in space
(interlocal or local), this event may reappear in the text in the form of an
infinitive construction or a subordinated clause. If the simple event is: “Jim
fell from the tree” (stratum 1.1). This sentence may appear as subordinated
to a sentential frame with the following verbs.
 NATURAL ONTOLOGIES AND SEMANTIC ROLES IN SENTENCES 191

I (John) saw Jim falling from the tree. (stratum 3.1)

I (John) saw that Jim fell from the tree. (strata 3.1/3.2)
I noticed that Jim fell from the tree. (strata 3.1/3.2)
I know that Jim fell from the tree. (stratum 3.2)
I believe that Jim fell from the tree. (stratum 3.2)
Jim’s falling from the tree. (stratum 4.2)
The speaker not only perceives, recalls and knows (believes) events and
actions; he may also imagine further events/actions, and consider possible
outcomes of current events/actions. Parallel to and ahead of current events,
alternative possibilities may be considered and possible outcomes may be
in the intentional domain of the speaker. This leads to further strata of
imagined realities, fantasies, etc.

NOTES

1 Translation by W.W.: “The multitude of forms makes up a realm of layers, whose

ranking is well known in its general shape: thing, animal, man, society”.
2 Translation by W.W.: “The common mind is not supported by a common conscience,
but only by individual consciences. This has not at all the same supporting power for it.
Thus its way of being remains in spite of all analyses of it a metaphysical and enigmatic
one”.
3 Translation by W.W.: “It (the mind) transcends the sphere of conscience by its commu-
nicability, by its transition from mind to mind, by the traditions handled from generation
to generation; in a similar way the closed circle of individual conscience does not take its
richness of content from itself but from the intellectual common ground of the life sphere,
into which it penetrates and to which it is adapted”.
4 Translation by W.W.: “This domain of being is the one of language and law, of custom
and political direction, of knowledge and art”.
5 As many of Hartmann’s papers and books were published during the Nazi-regime and
after the second world-war, a comparison with Cassirer would have to include the different
reactions to this political period. This would show the radical difference between the two
thinkers and could illustrate the recent history of German philosophy.
6 Translation by W.W.: “It presupposes the whole breadth of experience, as well that of
every day life and practical life as that of science. One may even add that it presupposes
the philosophical experience, i.e., the one which is inscribed in the historical process of
human thinking as a long series of trials, errors and self-corrections”.
7 In the more technical context of computational linguistics specifically in “Natural
Language Processing” (NLP) “ontologies” fulfil functions, like: organizing ‘world know-
ledge’, ‘organizing the world itself’, etc. (cf. Bateman, 1992).
8 Translation by W.W.: “Becoming is rather the general form of the real”.
9 Translation by W.W.: “And for just this reason the essence of the process is not con-
tained in the separation into stadiums. The stadiums are only short transitory stadiums,
momentary states; but although the states come and go, the process remains”.
192 WOLFGANG WILDGEN

10 Translation by W.W.: “As a state of conscience the process is an eminent form of

perceiving and enacting, and not just a form of penetrating understanding”.
11 The basic question remains if a proper ontology can be found by “accumulation” of
relevant criteria. In NLP-research three types of ontologies are commonly distinguished: a
(non-linguistic) conceptual ontology based on common sense or other criteria, an interface-
ontology, “an abstract organization underlying our use of grammar and lexis” (Bateman,
1992: 7), and a mixed ontology combining an abstract semantico-conceptual representation
of real-world knowledge that also functions as a semantics.
12 In lexicography content based dictionaries had already devised such ad hoc ontologies
which depend heavily on the decisions of lexicographers in a national tradition (cf. for
German Wehrle-Eggers, and Dowseiff).
13 A mathematically explicit analysis is given in Wildgen (1985: 208–212). The compacted
elliptical umbilic has a two-dimensional behaviour-space and consecutively shows (along a
linear path through the bifurcation set) the configurations: (A, B, I), (B, I, P). The attractor
B disappears and I (object) is caught by P. The underlying schema is that of giving: (A, I,
P).

REFERENCES

Barwise, Jon and Perry, John: 1984, Situations and Attitudes, Cambridge, MA: MIT Press.
Bateman, John A.: 1992), ‘The Theoretical Status of Ontologies in Natural Language Pro-
cessing’, in S. Preuß and B. Schmitz (eds.), Text Representation and Domain Modelling
– Ideas from Linguistics and AI, KIT-Report 97, Berlin: Technical University, pp. 50–99.
Bateman, John A.: 1995, ‘On the Relationship between Ontology Construction and Natural
Language: A Sociosemiotic View, Int. Journal of Human-Computer Studies 43, 929–
944.
Brandt, Per Aage: 1992, ‘La charpente modale du sens’, Pour une sémio-linguistique
morphogénétique et dynamique, Benjamins, Aarhus: Aarhus University Press.
Carnap, Rudolf: 1928/1961, Der logische Aufbau der Welt/Scheinprobleme in der Philo-
sophie, 2nd edn, 1961, Hamburg: Meiner.
Cassirer, Ernst: 1923–1929/1964ff, Philosophie der symbolischen Formen, Darmstadt:
Wissenschaftliche Buchgesellschaft.
Cassirer, Ernst: 1906-1950/1991, Das Erkentnisproblem in der Philosophie und Wis-
senschaft der Neueren Zeit, Vols. 1–4, Darmstadt: Wissenschaftliche Buchgesellschaft.
Dretske, Fred, I.: 1986, Knowledge and the Flow of Information, Cambridge,
MA/Bradford: MIT Press.
Fauconnier, Gilles: 1984, Espaces mentaux. Aspects de la construction du sens dans les
langues naturelles, Paris: Minuit.
Fauconnier, Gilles: 1999, Mappings in Thought and Language, Cambridge: Cambridge
University Press (reprint, first published 1997).
Fauconnier, Gilles and Turner, Mark: 2002, The Way We Think: Conceptual Blending and
the Mind’s Hidden Complexities, New York: Basic Books.
Fillmore, Charles: 1968, ‘The Case for Case’, in E. Bach and R. T. Harms (eds.), Universals
in Linguistic Theory, London: Holt, pp. 1–88.
Fillmore, Charles: 1977, ‘The Case for Case Reopened’, in P. Cole and J. M. Sadock (eds.),
Syntax and Semantics, Vol. 8, Grammatical Relations, pp. 559–581.
 NATURAL ONTOLOGIES AND SEMANTIC ROLES IN SENTENCES 193

Kelso, J. A. Scott: 1997, Dynamic Patterns. The Self-Organization of Brain and Behavior,
Cambridge/London: MIT Press.
Langacker, Ropnald: 1987, Foundations of Cognitive Grammar, Vol. 1, Theoretical
Prerequisites, Stanford: Stanford University Press.
Hartmann, Nicolai: 1933/1955, ‘Systematische Selbstdarstellung’, reprinted in Nicolai
Hartmann, 1955, pp. 1–51.
Hartmann, Nicolai: 1940/1955, ‘Neue Ontologie in Deutschland’, reprinted in Nicolai
Hartmann, 1955, pp. 52–89.
Hartmann, Nicolai: 1955, Kleine Schriften. Bd. 1: Abhandlungen zur systematischen
Philosophie, Berlin: de Gruyter.
Hartmann, Nicolai: 1964. Neue Wege zur Ontologie, Darmstadt: Wissenschaftliche
Buchgesellschaft (original edition 1947).
Lichter, Werner: 1964, Die Kategorialanalyse der Kausaldetermination. Eine kritische
Untersuchung zur Ontologie Nicolai Hartmanns, Bonn: Bouvier.
Marcus, Salomon: 1967, Introduction mathématique à la linguistique structurale, Paris:
Dunod.
Petitot-Cocorda, Jean : 1992, Physique du Sens. De la théorie des singularités aux
structures sémio-narratives, Paris: Editions du CNRS.
Wildgen, Wolfgang: 1981, ‘Archetypical Dynamics in Word Semantics: An Application of
Catastrophe Theory’, in Hans-Jürgen Eikmeyer and Hannes Rieser (eds.), Words, Worlds,
and Contexts, New Approaches to Word Semantics, Berlin: De Gruyter, pp. 234–296.
Wildgen, Wolfgang: 1982, Catastrophe Theoretic Semantics. An Elaboration and Applic-
ation of René Thom’s Theory, Amsterdam: Benjamins.
Wildgen, Wolfgang: 1985, Archetypensemantik. Grundlagen einer dynamischen Semantik
auf der Basis der Katastrophentheorie, Tübingen, Narr.
Wildgen, Wolfgang: 1994, Process, Image, and Meanings. A Realistic Model of the
Meanings of Sentences and Narrative Texts, Amsterdam, Benjamins.
Wildgen, Wolfgang: 1999, De la grammaire au discours. Une approche morphody-
namique, series: European Semiotics 1, Bern : Lang.
Wildgen, Wolfgang: 2001, ‘Kurt Lewin and the Rise of “Cognitive Sciences” in Germany:
Cassirer, Bühler, Reichenbach’, in Liliana Albertazzi (ed.), The Dawn of Cognit-
ive Science, Early European Contributors, series: Synthese, Dordrecht: Kluwer, pp.
299–332.
Wildgen, Wolfgang: in preparation for 2003, Evolutionary Semiotics, manuscript.