Review of Analogy as Structure and Process
| AUTHOR: Itkonen, Esa
TITLE: Analogy as Structure and Process
SUBTITLE: Approaches in linguistics, cognitive psychology and philosophy of
SERIES: Human Cognitive Processing 14
PUBLISHER: John Benjamins
Mareike Buss, Institute for Linguistics and Communication Studies (ISK),
RWTH Aachen University, Germany
In his book ''Analogy as structure and process'' Esa Itkonen presents a
critical reevaluation of the concept of analogy which has played a
prominent role in Western thought ever since Greek-Roman antiquity. He
discusses the historical and theoretical implications of this concept
mainly from a linguistic point of view, but also treats it from the
perspective of other disciplines such as philosophy and cognitive science.
His aim is ''to show [...] that analogy is THE central concept of language
and linguistics.'' (6) Moreover, he intends to ''authenticate the usefulness
of analogy as a methodological tool of contemporary research.'' (12)
The book contains a short introduction, four chapters with different
thematic focuses as well as a chapter with concluding remarks and, finally,
an appendix, where a formal approach to syntactic analogy is presented.
The first chapter ''The concept of analogy'' gives a comprehensive overview
over the different systematic aspects of analogy. Analogy as structure is
contrasted to analogy as process: while the former is defined as a static
relation between at least two systems, the latter is described as a dynamic
relation producing analogical structures. Even though the chapter is of a
mostly general character, many references are made to linguistics in order
to underline the relevance of the concept for the study of language.
In section 1.1 analogy is introduced as a metarelation: it is a relation on
an abstract level holding between at least two systems that, again, consist
of relations between their elements (in the following, 'at least two
systems' is referred to as 'two systems'). Analogy is, thus, based upon
structural similarity between two systems, while contiguity is the relation
between the elements of one system. Analogical relations may be
characterized further by specifying the kind of similarity-relation they
imply which can range from purely structural or formal similarity (e.g.
mathematical analogies) to structural~functional similarity (e.g.
anatomical analogies, where structural similarity is found to coincide with
functional similarity). Moreover, Itkonen differentiates two types of
analogy: static analogy comprises relations between given structures, while
dynamic analogy refers to those cognitive processes which static analogies
are established by.
After highlighting the context-dependency of analogy (1.2), a four-way
taxonomy of the relations between analogous systems is presented in section
1.3. This provides a basis for distinguishing between discovery,
invention/creation and imitation.
A distinction between two types of generalization is introduced in section
1.4. Generalization-1 describes the process of explaining a more abstract
phenomenon B by means of analogizing it to the more concrete phenomenon A.
Generalization-2, by contrast, consists in hypothesizing a common structure
X underlying both phenomena A and B.
In section 1.5 the relations of dynamic analogy to various types of
inference (induction, deduction, and abduction) are discussed. While
deduction obviously differs from analogy, the distinction between analogy
and induction is not so clear. Nevertheless, they are shown to be divergent
cognitive operations because the latter is based upon material similarity
between phenomena, while the former relies on structural similarity. Most
intricate is the relation of analogy and abduction, since abduction always
implies some kind of analogical reasoning, while dynamic analogy must
basically be conceived of as an abductive process.
In section 1.6 analogy is contrasted to metaphor and cognitive blending. An
analysis of the well-known metaphor ''TIME IS SPACE'' serves to demonstrate
that linguistic metaphors generally contain several analogies, linked to
one another in a kind of hermeneutic cycle. Metaphor as well as cognitive
blending is defined as ''a subtype of analogy, or 'an analogy with added
constraints': all metaphors are analogies, but not all analogies are
Subsequently, Itkonen discusses the issue of psychological adequacy of
general linguistic descriptions and of analogy-based linguistic
descriptions (1.7). With regard to synchronic linguistics, he advocates a
clear division of labor between so-called autonomous linguistics and
psycholinguistics, the former setting up hypothetical linguistic rules and
the latter verifying experimentally their psychological adequacy. In a
diachronic perspective, however, one has to rely upon other than the
experimental methods of psycholinguistics to prove the psychological
reality of theoretical constructs.
In section 1.8 human behavior is subdivided into instrumental and
institutional behavior. Instrumental behavior comprises all forms of
rational actions that presuppose a (free) choice between different means to
achieve a certain goal; it is subject to rational analysis and explanation.
By contrast, institutional or rule-governed behavior cannot be rationally
explained, because the individual normally acts according to already
existing and well-established rules and not according to rational choices.
In synchronic linguistics the difference between instrumental and
institutional behavior is instantiated by the difference between pragmatics
(i.e. free rational linguistic behavior) on the one hand and morpho-syntax
and semantics (i.e. rule-governed institutional behavior) on the other.
Section 1.9 deals with the problem of competing analogical interpretations
of a given phenomenon. Analogical ambiguity results from the possibility to
choose more than one adequate analogical model for the interpretation of a
Up to this point the first chapter was entirely dedicated to highlighting
the ubiquity of analogy in cognitive operations. In the last section 1.10,
however, the author points out that there are necessarily 'limits of
analogy': ''Analogy must (ultimately) rest on something non-analogical;
otherwise there would be infinite regress.'' (63) In other words: even if we
assign an absolutely central role to analogy in human intellectual life,
supposing that it is involved in all types of invention and learning,
analogical operations must be grounded on something else than analogy, e.g.
innate basic concepts. Furthermore, they must be susceptible to be
contrasted to non-analogical operations, e.g. induction, deduction,
falsification, experimentation, etc.
The second chapter ''Analogy inside linguistics'' deals with the
epistemological and methodological relevance of analogy in the different
linguistic disciplines. Before turning to a detailed analysis of the role
of analogy in linguistics, Itkonen confronts in section 2.2 the manifold
objections against analogy that have been stated/expressed in modern
linguistics. It is mainly in generativist linguistics that analogy has been
explicitly banned as a relevant concept of linguistic theory. The author
shows, however, that – in spite of the explicit exclusion of analogy –
generativism is deeply indebted to the concept of analogy at the level of
both theoretical and empirical research.
Section 2.3 outlines the importance of analogy as one of the foundational
notions of structural phonology. N. Trubetzkoy introduced this notion as
the organizing principle of phonological systems: phonemes are defined as
bundles of distinctive features that are constituted by their similarities
and differences with regard to the other elements of a given system.
Itkonen points out that there are two types of relations between phonemes:
they either stand in proportional opposition to each other (as in p:b = t:d
= k:g) or in isolated opposition (as in k:s). Proportional oppositions are
based on analogy and contribute to the coherence and, thus, to the
stability of the whole system, while isolated oppositions do not have any
relevance for the system.
In section 2.4 analogy is shown to play an equally important role in the
organization of the morphological system of a given language. The
Neogrammarian linguist H. Paul describes language as a system of word
classes that are based on form and/or meaning associations. He
distinguishes two types of classes: so-called 'material' and 'formal
groups'. The former comprise all inflected instances of a single word, e.g.
the complete declension of the Latin word 'arbor'. The latter, in contrast,
are grammatical paradigms comprising all the instances of a whole word
class in a certain case, e.g. all Latin nouns in the dative case. Whereas
material groups are simply based on material identity, formal groups are
organized by analogical relations between their elements that guarantee the
coherence and the learnability of the system: ''Whatever else the
inflectional morphology of a given language may be, it is certainly a
system held together by crisscrossing analogies [...]. Otherwise, it would
be simply impossible to learn and memorize such huge amounts of data'' (83).
As for the role of analogy in syntactic theory, things are a little bit
more complicated. This is why Itkonen briefly analyzes in section 2.5 three
different approaches to syntax, the Arabic tradition, generativist
linguistics and cognitive linguistics, which all reveal a different account
of the theoretical importance of analogy. To begin with the Arabic
tradition of grammaticography, one of its central methodological notions is
'qiyás' which translates literally as 'measuring' and consists in
discovering syntactic patterns by means of analogical reasoning. Analogy
thus counts here as a central principle of describing syntactic structures.
Generativist linguistics, on the contrary, exhibits an ambivalent attitude
to analogy: while explicitly refuting this notion, generativists implicitly
rely upon analogy when explaining even the most basic and simplest
examples. What is true for the theoretical explanations also holds for the
construction of the theory on a more general level. Itkonen points out that
analogical reasoning plays a crucial role in the whole theoretical
architecture, e.g. the analogy between deep structure and surface
structure, the analogies established between different lexical categories,
etc. In cognitive linguistics, more precisely in construction grammar,
analogy finally plays again a major role with regard to both the alleged
underlying linguistic structures and their linguistic descriptions. Basic
verb-meanings, for instance, are thought to be extended to non-basic
verb-meanings via analogy; the same holds for the extension of basic to
In the domain of semantics (2.6), analogy has become an issue in the
discussion of linguistic iconicity. Iconicity of language is a plain
exemplification of analogy, since it implies two important theoretical
claims: (1) there are structural similarities between non-linguistic
(world/concepts) and linguistic entities, (2) the similarity-relation is
directed from non-linguistic to linguistic entities, i.e. from
world/concepts to language.
Section 2.7 deals with diachronic linguistics, which has traditionally made
extensive use of the notion of analogy, in order to explain language
change. This is also true with regard to the study of grammaticalization, a
diachronic two-stage process consisting first in the reanalysis of a given
linguistic unit and then its (analogical) extension. It has always been
recognized that analogy is the driving force behind the extension of a
given pattern. By confronting grammaticalization with the process of
abduction (cf. 1.5), Itkonen shows, however, that analogy is important for
both reanalysis and extension.
In section 2.8 the ''analogy between oral languages and sign languages''
(113) is discussed. Both oral and sign languages are said to be strictly
symbolic in nature, even though they manifest a different degree of
iconicity, sign languages being more iconic than oral languages. An
exemplary analysis of the sentence-meaning 'N1 gave/showed N2 to N3' in
Finnish Sign Language, Japanese, English, Latin, Finnish, West-Greenlandic,
Wari', Swahili, Yagua, and Yimas serves to demonstrate that the analogy
between oral and sign languages ''exists as a matter of fact'' (117). Chapter
2 is concluded with some remarks concerning analogy and language acquisition.
The third chapter ''Analogy and/or overlap between language and other
cognitive domains'' deals with the pervasive structural similarities, i.e.
analogies between language and other domains like images/pictures, vision,
music, and logic. On the one hand, the chapter aims at providing arguments
against the hypothesis of the 'modularity of the mind' (N. Chomsky, J.
Fodor). On the other hand, the presented findings are to be considered as a
contribution to the long-lasting debate in semiotics and philosophy of
language about the relations between language and other symbolic systems. A
central claim of the modularity-hypothesis is the modular encapsulation of
language in the mind/brain with regard to other cognitive domains. Itkonen
demonstrates, however, that this claim is untenable because there are
evident structural similarities between language and world (= iconicity) as
well as between language and other cognitive domains which, moreover, even
In section 3.2 the notion of iconicity that had previously been introduced
(cf. 2.6) is discussed with regard to the modularity-hypothesis. Iconicity
is per definition opposed to the concept of modularity and the alleged
encapsulation of cognitive capacities, because an iconic relation
necessarily involves two terms or entities, e.g. language and world. If the
idea of iconicity could be proven to be true, the modularity-hypothesis
would be proven to be wrong. But iconicity may also shed new light on the
classical semiotic distinction between verbal and pictorial representation:
''the assumption of dual coding ('propositional vs. pictorial') [...] is
misconceived, because propositional is pictorial: language imitates, and is
a picture of, reality and assuming that the structure of thought imitates
the structure of language, the former is a 'picture' of the latter.'' (136)
In section 3.3 it is shown that language and vision rely on analogous
cognitive processes, inasmuch as linguistic and visual perception are
holistically organized by 'Gestalts'. The analogy of language and vision,
more precisely ''the ultimate unity of all types of perceptual experience''
(137) is again interpreted as an argument against the alleged modularity of
The same holds for the analogy of language and music, which in section 3.4
is first illustrated with regard to oral languages, e.g. by pointing to
structural similarities of the respective syntactic constituent structure.
Subsequently, the analogy between language and music is also extended to
sign languages. These are, for instance, susceptible to a melodious use,
which is analogous to humming in oral languages. However, Itkonen does not
only claim that language and music synchronically bear upon common
cognitive resources, he also suggests that they most probably have a common
evolutionary origin. In section 3.5 he finally exhibits that language and
logic have a common origin, more precisely that (formal) logic is
originally derived from 'ordinary language'.
The fourth chapter ''Analogy (mainly) outside linguistics'' illustrates the
central role analogy generally plays in human thinking. There are three
main areas of interest for examining the role of analogy outside
linguistics: commonsensical reasoning, mythological reasoning, and
scientific reasoning. While common sense thinking is not treated, analogy
in mythological and scientific reasoning is discussed in more detail. It is
demonstrated that the allegedly clear line of demarcation between myth and
science is getting blurred under the perspective of analogy.
Section 4.2 deals with the importance of analogy in myth and cosmology.
Anthropological research has shown that binary classifications universally
characterize mythological reasoning. Since binary classifications relate
different entities in conceptual pairs according to structural
similarities, they are chief examples of analogical reasoning. This kind of
analogical conceptualization of the world is illustrated with the
Tolkaappiyam, a Tamil grammar dating from 500-0 BCE. It does not only
comprise grammatical descriptions, but also a poetological treatise. The
treatise analogically relates language, poetry and social behavior to one
another within the cosmology of the ancient Tamil culture.
Section 4.3 provides a whole range of examples to corroborate the
hypothesis that analogy is a crucial mechanism in the process of scientific
discovery. In the first subsection examples from the history of the natural
sciences, philosophy, and technology illustrate the importance of analogy
for scientific reasoning in general. A well-known analogy is the clockwork
analogy that served J. Kepler to elucidate the functioning of heaven, while
R. Descartes used it to explain the structure of the physical world. The
second subsection discusses some famous linguistic analogies, for instance
the analogical conception of language as an instrument (Plato, L.
Wittgenstein) or as a game (F. de Saussure, L. Wittgenstein). The author
concludes the chapter with some remarks as to the relevance of his findings
for the philosophy of science.
In his concluding remarks the author restates his main theoretical claim:
''thinking is basically analogical.'' (199) The appendix, finally, contains
the PROLOG program which formalizes the analogical inferences underlying
the syntactic examples of section 2.5.
Esa Itkonen's approach to analogy is very 'continental' in method and
style. He investigates analogy in a systematic and historical perspective,
i.e. by making reference to a whole range of major Western philosophers and
linguists. I think that such a historical perspective is a fruitful way to
reevaluate the foundational problems of theoretical linguistics. Retracing
the history of a theoretical problem or concept may help to elude old and
well-known fallacies and, hence, to restate the problem more clearly in
modern terms. Still, it is mandatory not to indulge in merely listing all
those thinkers, who have already treated a certain question. The historical
perspective must rather be embedded into an systematic account of the
theoretical problem. In his book Itkonen aptly combines the systematic and
the historical approach and, therefore, provides a fresh and stimulating
survey of the concept of analogy.
One of the most interesting aspects is the distinction between 'analogy as
structure' and 'analogy as process', which is established in the first
chapter. Itkonen very convincingly argues that it leads to important
theoretical and methodological consequences. He shows, for instance, that
some major methodological flaws in generativist linguistics are due to the
lack of recognizing this distinction (72). In my opinion, the
structure-process-distinction applies also to other linguistic phenomena
and, hence, opens up some perspective for future theoretical research.
However, I would also like to add some critical remarks. The first concerns
the theoretical and terminological delimitation of the treated phenomena,
which are not always rigorously distinguished. Let's take, for instance,
the difference between analogy and similarity. On the one hand, analogy is
defined as a relation that is established between two entities on the basis
of structural similarities (1.1). On the other hand, analogy is sometimes
simply equated to (structural) similarity, e.g. in chapters 3 and 4. In my
opinion, the difference between analogy and similarity should have been
exposed more clearly. This could have been achieved, for instance, if it
had explicitly been related to the distinction between 'dynamic' and
A similar remark holds for the distinction between analogy, metaphor, and
model. During the last decades these terms have been exhaustively discussed
in linguistics (e.g. Ortony 1993), in philosophy of language (e.g. Black
1962, Ricoeur 1979), and even in philosophy of science (e.g. Hesse 1980).
Much effort has been spent on their systematic distinction. Still, the
distinction of analogy and metaphor, outlined in section 1.6, remains on a
rather superficial level and is sometimes even completely abandoned in
terms like ''analogy-as-metaphor'' (38). This is also true for the
distinction between analogy and model (185, 191). Such terminological
imprecisions sometimes contribute to the impression that the single
chapters of the book are rather loosely connected to one another.
Finally, the text sometimes lacks theoretical justification as to the
choice of theorists, whose positions are discussed. Although Itkonen has
provided an all in all convincing combination of systematic and historical
argumentation, this occasionally may convey an impression of arbitrariness.
Furthermore, a few accounts of philosophical ideas either reveal a rather
superficial analysis (e.g. the reference to Wittgenstein's 'picture
theory', 115, 145) or they do not contribute to the central arguments of
the text (e.g. the comment on Heidegger, note 12, 222).
The proofreading has been very good, since the book contains only a few
typos. Yet, there are some inconsistencies between the references in the
text and the bibliography, e.g. Wittgenstein (1969 ) is not listed in
On the whole, the book is an absolutely inspiring and interesting read. It
gives a broad overview over the concept of analogy and its theoretical and
methodological implications. Itkonen persuasively shows that analogy is a
ubiquitous phenomenon in human culture, particularly in language. It has
wrongly been neglected by theoretical linguistics during the last decades.
The book deals rather critically with regard to generativist linguistics,
while it favors functional-typological approaches. However, it could be of
interest for linguists from all theoretical schools, who are open to
reconsidering certain theoretical problems and concepts in a broader
philosophical perspective. Undoubtedly, linguists studying language from a
cultural perspective will profit the most from Itkonen's reflections.
Black, Max (1962) Models and Metaphors: Studies in language and philosophy.
Ithaca, N.Y.: Cornell University Press.
Hesse, Mary (1980) Revolutions and Reconstructions in the Philosophy of
Science. Brighton: Harvester.
Ortony, Andrew, ed. (1993) Metaphor and Thought. Cambridge: Cambridge
Ricœur, Paul (1979) The Rule of Metaphor. Toronto: Toronto University Press.
| ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Mareike Buss is currently working as a teaching and research assistant at
the Institute for Linguistics and Communication Studies (ISK), RWTH Aachen
University, Germany. She is currently finishing her PhD thesis about
'iteration' as the central semiotic mechanism governing the interaction of
language system and language use. Her research is concerned with
usage-based models of language, functional theories of grammar,
metaphorology and semiotics as well as the historiography of (modern)