prototype theory prototype semantics
TRANSCRIPT
Prototype Theory and Prototype Semantics
by
Victoria Mears
Philipps University Marburg
1 Introduction 2
Contents
1 Introduction......................................................................................................3
2 Prototype Theory and Prototype Semantics .................................................4
2.1 The Origins of the Prototype Approach ............................................................4
2.1.1 Aristotle: The Classical Approach 4
2.1.2 Gottlob Frege: The Defining-Attribute View 5
2.1.3 Ludwig Wittgenstein: Family Resemblance 6
2.2 Prototype Theory I.............................................................................................6
2.2.1 On the Way to the Prototype Theory 6
2.2.2 The Standard Prototype Theory 7
2.2.3 Hierarchic Structure: Basic Level Categories 10
2.2.4 Problematic Aspects of the Standard Prototype Theory 11
2.3 Prototype Theory II .........................................................................................12
2.3.1 Dirk Geeraerts: Prototypical Prototypicality 12
2.3.2 Ronald Langacker: Schema 13
2.3.3 George Lakoff: Idealized Cognitive Model (ICM) 14
3 Conclusion ......................................................................................................15
4 References.......................................................................................................16
5 Appendix.........................................................................................................17
1 Introduction 3
1 Introduction
The world consists of a limitless variety of objects and concepts. How does
humankind reduce this vast input of constant information into manageable
portions? And how does language reflect this mental input?
The answer lies within mankind’s nature to cope with his environment: in the
cognitive ability of categorization. Categorization presupposes that objects and
concepts are comprised of attributes which can be put in opposition to each other.
Hence, a distinction between objects and their cognitive reference point can be
drawn. Thus, for example, human beings are primarily categorized into male and
female. Without this classification in terms of binary opposition, mankind would
not be able to grasp its environment with its constant massive input. As
classification and its logically connected categorization is a feature of human
cognition, it is naturally reflected in language. Distinctive objects receive
distinctive words.
But how come we call a dog a ‘dog’? Why do we label a Chihuahua, a
bulldog and an Alsatian with the lexeme DOG? The first animal is a tiny, yapping
lap-dog. The second is associated with defense of property due to its aggression
and viciousness, and the third animal seems to be a ‘better’ example of a dog,
doesn’t it? But where does the definition of DOG end and that of WOLF begin?
Apparently, categorization is not as clear-cut and binary as it seems at first glance.
This paper will deal with the basic theory of prototypes and prototype
semantics, its features and implications. After a short introduction to the classical
approach and the defining-attribute theory, the standard prototype theory,
including the basic level structure, will be presented and analyzed. The
problematic aspects with the prototype will be shown, as well. Furthermore, the
expansion of the theory as proposed by Lakoff, Langacker and Geeraerts will be
depicted very briefly. Due to the wide scope of the theory and the limited number
of pages of this paper, merely a short overview can be given, which hopes to set
down the basic notions of the prototype theory.
2 Prototype Theory and Prototype Semantics 4
2 Prototype Theory and Prototype
Semantics
In order to understand the prototype theory and its extension, one must first take a
look at its origins.
2.1 The Origins of the Prototype Approach
Basically, the ideas of Aristotle, Gottlob Frege and Ludwig Wittgenstein
contributed markedly to the prototype approach which finally led to the
establishing of the prototype theory.
2.1.1 Aristotle: The Classical Approach
Most linguists ascribe the so-called classical approach to categorization to
Aristotle, who set down his definitions of concepts in his work Metaphysics in
roughly 350 BC. Up until the late 20th century, his notions were taken for granted
in the disciplines of psychology, philosophy and linguistics. The Aristotelian
definition states that a category is defined by a set of necessary conditions which
are together sufficient and determine its membership. In other words, Aristotle
distinguishes between the ‘essence’ of a thing and its ‘accidents’. ‘Accidents’ do
not determine the thing as a whole and are thus not necessary. However, the
‘essence’ defines the thing in its true status. The essence is “[...] all parts
immanent in things which define and indicate their individuality, and whose
destruction causes the destruction of the whole” (Metaphysics 5.8.3 in Taylor
1995: 22). As an example one may take the concept GIRL: X is human, female
and young. All these criteria are necessary, and if one is missing, then X is not a
girl. Remove, e.g., young and X is a woman. This phenomenon, in turn, means
that these attributes are jointly sufficient for the definition of GIRL.
Various assumptions underlie Aristotle’s concept: firstly, categorization relies
on a set of necessary and sufficient conditions which are in every case
2 Prototype Theory and Prototype Semantics 5
individually necessary and jointly sufficient. Secondly, the features always appear
in a binary relationship, e.g., if X is a girl, X has to be female. In the case that X is
male, the definition collapses. So, these conditions are a matter of yes-or-no,
which – in turn – defines the category membership. Thirdly, according to the
classical approach, a category has a fixed and clear-cut boundary. This arises from
the basis of the theory which states that a word meaning can be defined exactly
and thus, a clear distinction to other lexical meanings may be drawn. But how can
natural categories, such as colors, numbers or shapes, or abstract concepts, e.g.
hate, beauty or freedom, be defined? Fourthly, this notion of classification implies
that all members of a category have equal status as full members. Hence, if all
criteria of the definition of membership are fulfilled, the entity belongs to the
category. But then why are some members of a category instinctively thought to
be better examples, whereas others are considered as marginal or borderline
cases? Aristotle’s theory raises problematic aspects and does not satisfy many
questions.
2.1.2 Gottlob Frege: The Defining-Attribute View
Another theory that follows the classical approach strongly is that of the German
logician Gottlob Frege and his followers, namely the ‘defining-attribute theory’,
which was developed in the mid 20th century. In summary, this concept maintains
that a category has defining features, whereby a division into the ‘intension’ and
‘extension’ is drawn. The ‘intension’ resembles the set of attributes which mark
the definition of a member of the category; the ‘extension’ is the set of entities
that comprises the members. Michael W. Eysenck and Mark T. Keane give the
example of the concept BACHELOR: in this case, the defining attributes are
male, single and adult. For X to be a bachelor, these three attributes are necessary
and sufficient. The complete set of every single bachelor in the world, ranging
from the Pope to one’s neighbor, is the ‘extension’ of the notion (Eysenck/Keane
2000: 285).
However, the same problems which arise concerning Aristotle’s classical
approach apply to Frege’s view, as well.
2 Prototype Theory and Prototype Semantics 6
2.1.3 Ludwig Wittgenstein: Family Resemblance
In his work Philosophical Investigations (1953), the philosopher Ludwig
Wittgenstein proposed the concept of ‘family resemblance’. He disassociated
himself from the classical approach by claiming that members of a category do
not share common attributes – as surmised by Aristotle – and that they instead
resemble one another in a way that link them to the category. Some members may
share no features but fulfil linking criteria to other members. In short, members
are linked by a network of similarities. These concurrences can be formulated in
mathematics as AB, BC, CD, etc. Wittgenstein substantiates his theory with the
following example of the category GAME:
[...] Board-games, card-games, ball-games, Olympic games, and so on. What iscommon to them all?-Don’t say: “There must be something common, or they wouldnot be called ‘games’” – but look and see whether there is anything common to all. –For if you look at them you will not see something that is common to all, butsimilarities, relationships, and a whole series of them at that. [...] And we can gothrough the many, many other groups of games in the same way; can see howsimilarities crop up and disappear. (Wittgenstein 1963: §66, pp.31e-32e)
Not only does the example of the category GAME prove that some categories
seem to be fuzzy and difficult to define, it also shows that category members are
learnt by similar exemplars and not on the basis of criterial features. This
discovery formed the foundation for the prototype theory.
2.2 Prototype Theory I
2.2.1 On the Way to the Prototype Theory
To account for the deficiencies of the classical approach and the ‘defining-
attribute theory’, the prototype theory was proposed.
In 1969, the anthropologists Brent Berlin and Paul Kay published their
findings on basic color terms. They examined 98 languages (20 of them in detail)
in order to investigate whether color categories were arbitrary or not, an
assumption which had been long-believed and supposedly proved in the 1950s
and 1960s. Using a set of over 300 chips, native speakers of various languages
were asked to first label chips with an one-morpheme term and then choose the
2 Prototype Theory and Prototype Semantics 7
‘best’ examples of a color term. The result of their experiments was revolutionary:
the basic color categories are defined by so-called focal colors1 which are
reference points in a color spectrum. Focal colors were not only chosen by
speakers of the same language community, but were also a cross-linguistic
phenomenon. Contrary to the foci, the names of colors between the basic colors,
such as the English term magenta or the German Indigo or Ocker, do not have
clear boundaries – neither between speakers of the same language nor between
speakers of different languages.
Eleanor Rosch, a psychologist at the University of California in Berkeley,
picked up Berlin and Kay’s results and extended the tests on focal colors with
experiments with 3 to 4-year-old children. Her research produced the following
discoveries:
1. Focal colors stand out perceptually more than non-focal ones.
2. Focal colors are taken up more rapidly into the short and long-term memory.
3. Children are able to produce names for the focal colors more quickly and can
store these names from an early stage on.
These results lead Rosch to replace the term foci by that of prototypes. A
prototype represents the best exemplar in a category.
2.2.2 The Standard Prototype Theory
Eleanor Rosch’s research on focal colors lead her to examine other categories:
Although the present research dealt only with the domain of colors, it raisesquestions about semantic reference in other domains. [...] However, much actuallearning of semantic reference, particularly in perceptual domains, may occurthrough generalization from focal exemplars – a process made reasonable, if notproved, by the present study of a possible basis for the learning of color names.(Heider2 1971: 455 in Mangasser-Wahl 2000: 18).
In this quotation, it becomes apparent that the prototype serves as a referent point
for categorization, as it does not define the meaning of its member, but instead
evokes a cognitive picture associated with the term itself. Her most famous
experiment involved more than 200 psychology students who were requested to
_________________1 This term was coined by Berlin and Kay and is also called foci.2 Eleanor Rosch had the maiden name Heider.
2 Prototype Theory and Prototype Semantics 8
formulate the mental pictures words evoked when referring to categories. Rosch
began the test with the following introduction:Let’s take the word red as an example. Close your eyes and imagine a true red. Nowimagine an orangish red ... imagine a purple red. Although you might still name theorange red or the purple red with the term red, they are not as good examples of red... as the clear ‘true’ red. In short, some reds are redder than others. (Rosch 1975:198in Aitchison 1987:53)3
The students then had to judge the ‘goodness’ or ‘typicality’ of category
members of, all in all, ten categories: BIRD, FRUIT, VEHICLE, VEGETABLE,
SPORT, TOOL, TOY, FURNITURE, WEAPON and CLOTHING. The rating
scale ranged from 1 point, which meant a very good example, to seven points,
which stood for the worst example in the list or for no example at all. The
prototypes are represented by the lowest average value. Friedrich Ungerer and
Hans-Jörg Schmidt depict a selection of Rosch’s results in their work An
Introduction to Cognitive Linguistics (1996: 13).4
As we see in the diagram and the picture of the rankings of birds5, the
interviewees rated robin the best exemplar of a bird, whereas ostrich and penguin
received very low ratings. Hence, it becomes apparent that the members of the
category BIRD do not have equal status, rather that the category members are
structured around a prototype. In addition, the category has an internal structure
with a prevailing hierarchy or degree of membership which is brought about by
resemblance to the prototype6. All members of the category BIRD have in
common a beak and the ability to lay eggs as minimal biological attributes7. For
most people, however, ‘proper’ birds should be light-weighted, have feathers, be
able to sing and, most significantly, be able to fly. Hence, ostriches and penguins,
both birds unable to fly, tend to be considered as borderline cases, as atypical
birds. None of these features are essential for distinction from other categories.
Hence, a duck-billed platypus lays eggs, can swim and has a beak similar to that
_________________3 In retrospect, psychologists assume that prototypicality of focal colors has to do with the neurology of color
perception, i.e., that, apart from focal colors, spatial orientation and geometrical forms in their ‘true’forms (e.g. circle, square, verticality, etc.) are more salient than deviant forms and thus markedly form theprototypes.
4 Please look at the diagram in the appendix, figure 1.5 Figure 2 of the appendix.6 Wittgenstein’s theory is useful as a supplement to the prototype theory. Figure 4 depicts the family
resemblance for the category BIRD.7 The attributes of BIRD are shown in the appendix, figure 3.
2 Prototype Theory and Prototype Semantics 9
of a duck. Even though it has many characteristics of a bird, it belongs to the
family of monotremes, which includes porcupines and ant-eaters – animals that
seem to share no features whatsoever. In this example it becomes apparent that
categories are fuzzy as the borderline between the categories is not precise but
instead flexible, i.e. new members may be accepted into the category.
Of course, these ‘goodness-of-exemplar’ ratings8 are dependent on the
cultural background of the interviewee. So, for example, an Arab will probably
rate fig the prototype of FRUIT, whereas it will receive a low rating for a
European who, most likely, considers apple or orange the best example of a fruit.
One assumes that the degree of membership may have to do with the higher
frequency of prototypes in our environment. However, Eleanor Rosch denied this
explanation of more frequent occurrence and put this discovery down to the order
in which we learn members of a category in our childhood.
In these various prototype tests, psychologists found out several properties in
labeling the prototypes. So, for example, subjects took less time verifying the
prototype than the marginal members: saying “yes” to the statement “an ostrich is
a bird” took by far longer than to “a robin is a bird”. In addition, when the
subjects were required to name an exemplar of a category, they usually mentioned
the prototype first.
The prototype theory is applicable beyond physical objects and can also be
extended to, e.g., verbal categories. The linguist Steve Pulman examined the
GOE-rates of the verbs to kill, to speak and to walk inter alia.9 His results show
that to murder represents the prototype of the category TO KILL, whereas to
commit suicide was considered a less typical example by most subjects.
Language itself possesses a mechanism for expressing various category
memberships. This mechanism was termed ‘hedges’ by the professor of cognitive
linguistics George Lakoff (1972). Hedges include sentence adjuncts, such as
strictly speaking, conjunctions like in that, modifiers, e.g., so-called, and
graphological forms, such as inverted commas in a ‘honest’ politician to express
_________________8 ‘Goodness-of-exemplar’ is abbreviated as GOE for sake of simplification.9 The results are presented in the appendix, figure 5.
2 Prototype Theory and Prototype Semantics 10
degree of category membership. One can mention Rosch’s bird example to
illustrate how language users apply hedges. Here are a few examples:
1. A robin is a bird par excellence.
2. A penguin is a kind of bird.
3. Strictly speaking, an ostrich is a bird.
2.2.3 Hierarchic Structure: Basic Level Categories
Georges Kleiber differentiates in his book Prototypensemantik: Eine Einführung
(1993) between the ‘horizontal’ and ‘vertical dimension’. He ascribes categories
and their prototypes to the horizontal dimension, the internal structure of
categories, and the so-called basic levels to the latter, which describes the
intercategorial structure. In Rosch’s words, there are “two axes of categorization”.
According to the principle of the vertical axis, there are different levels within a
category which have a hierarchic structure. Rosch and her associates classify this
dimension into three levels of abstraction: the superordinate, the basic and the
subordinate level. From the superordinate to the subordinate level, each category
has the features of the dominating category in addition to one or more
distinguishing attributes. One can take the lexeme dog as a simple example:
ANIMAL (superordinate level) > DOG (basic level) > CHIHUAHUA
(subordinate level).
The basic level is the dimension at which concepts possess the most
distinctive attributes and hence combines both informativeness and
distinctiveness, thus being cognitively highly economic. The more specific a
concept or object is, the more information it holds. Furthermore, the features of
the basic level have a high cue validity, which means that they are shared by a
high proportion of members and a low amount of non-members.
In psychological experiments, pictures of objects are categorized more
quickly at the basic level than at the other two levels. Furthermore, one can draw a
picture of a category’s basic level, but not of a superordinate level. Everybody can
draw a picture of a cat, but how does one depict the term feline? Linguistically,
basic level terms predominate in adult communication, are acquired by children as
2 Prototype Theory and Prototype Semantics 11
first category levels, and mostly have short and simple names, i.e., they are often
monomorphemic.
In contrast, subordinate levels are often expressed in composites, such as
rocking chair or marble cake. Superordinate terms sometimes show up deviant
forms in some way, such as the category FURNITURE, as it is an uncountable
noun. In other cases, the term above the basic level is just missing in a language,
e.g., a superordinate color term in English.
2.2.4 Problematic Aspects of the Standard Prototype Theory
Although the prototype theory brought great insight into cognitive organization of
categories, it still shows many unsolved problems. First of all, no category
boundary is given in the theory. However, is this not the main function of
categorization, namely to distinguish one object from another? How can one form
distinction without clear-cut boundaries? The classical approach offered these
boundaries but left no allowance for any internal structure. In addition, categories
with exact boundaries and binary membership do possess best examples: in the
category ODD NUMBER, for example, there are no marginal cases but rather two
defining conditions.10 But 3 or1 is definitely the preferred example in the
category, rather than, say, 24 478 315.
As discussed in chapter 2.2.2, the prototype is a full member of a category,
contrary to other members which have a degree of membership according to their
resemblance to the ‘ideal’ member, the prototype. But as resemblance or
similarity is a consequence of comparison, can one actually compare, e.g., lamp to
the prototypical chair in the category FURNITURE? Hence, the family
resemblance theory is useful for some categories but not for others. In connection
with resemblance, there are two problems concerning similarity: similarity is also
structured in degrees, i.e. something is more or less similar to something else, and
judgements of similarity can only be subjective, not objective or generally
specifiable.
_________________10 An odd number must be a natural number other than 0 and not divisible by 2 without a remainder.
2 Prototype Theory and Prototype Semantics 12
Another problem with the prototype theory is naming and ordering the
features of the prototype. Some characteristics are prior to others, such as ability
to fly in the category BIRD, but are catches worms and eats crumbs prototypical
features, as well? Probably not.
Furthermore, there is no end to encyclopedic knowledge in categorization.
Thus, a biologist, for example, will not have any difficulty in distinguishing the
categories BIRD and BAT. In consequence, the biologist will not find the
boundaries fuzzy in comparison to, say, a literary critic.11
These points mentioned are the main points of criticism - there are many
others which apply to specific subfeatures of the prototype theory.
2.3 Prototype Theory II
As the prototype theory shows the deficiencies presented above, the theory has
been extended and modified in recent years. A few novelties shall be presented
briefly.
2.3.1 Dirk Geeraerts: Prototypical Prototypicality
Dirk Geeraerts, a Dutch linguistics professor, claims that prototypicality is a
prototypical term in itself and thus tries to explain why the notion of prototype
does not apply to all categories. He combines different prototypical features to
base his theory of ‘reflexive prototypicality’. Firstly, he differentiates between
monocentric and polycentric categories. In the first case, the category refers to
different types of referents which are coupled with intuitive univocality. Thus, he
declares the category BIRD, for example, to be monocentric as it refers to robin,
ostrich and eagle without having several meanings, i.e., being polycentric.
Contrary to monocentrality, a clustering of several overlapping meanings leads to
a polycentrally structured category. Geeraerts uses the example of the Dutch
adjective vers, which means more or less fresh in English, for illustration.
_________________11 Here, the theory of folk and expert categories can be applied. Folk categories, which are based on
prototypes, rely on the way people perceive objects in their environment, whereas expert categories aregenerally conform with the classical approach. With the help of expert knowledge, boundaries can bedrawn.
2 Prototype Theory and Prototype Semantics 13
Michaela Zitzen lists the three meanings in her essay “On the Efficiency of
Prototype Theoretical Semantics”:
1. vers = new, novel, recent and therefore optimal (e.g. consumption of bread,
fruit)
2. vers = novel, recent (e.g. news and information)
3. vers = in an optimal condition, pure untained (e.g. air)
One realizes that Wittgenstein’s idea of family resemblance plays an important
role in Geeraert’s theory, as meaning 1 contains both meaning 2 and 3. The
monoreferential concept of the category becomes multireferential by this
extension.
2.3.2 Ronald Langacker: Schema
Similar to Geeraert’s version of polysemy, the cognitive linguist Ronald
Langacker extends the prototype theory by means of a ‘schema’, which he defines
as following:A schema, by contrast, is an abstract characterization that is fully compatible withall the members of the category it defines (so membership is not a matter of degree);it is an integrated structure that embodies the commonality of its members, whichare conceptions of greater specificity and detail that elaborate the schema incontrasting ways. (Langacker 1987: 371 in Taylor 1995: 66).
For example, the category TREE exemplifies Langacker’s schema:
TREE1 = large, deciduous leafed plants, which serve as the prototype
TREE2 = TREE1 plus pine trees
TREE3 = TREE2 plus palms (Taylor 1995: 66)
A more abstract extension is also possible, hence TREE2 could be expanded
metaphorically, e.g., with phrase structure tree, genealogical tree, tree of life, etc.
The basic problem with the theory of ‘schemas’ is the lack of restriction due to the
fact that it is too general. Additionally, there is a degree of membership, despite
Langacker’s negation, which is visible in the increasing schematic abstractness.
2 Prototype Theory and Prototype Semantics 14
2.3.3 George Lakoff: Idealized Cognitive Model (ICM)
George Lakoff (1987) created a prototypical model called the ‘Idealized Cognitive
Model’, which is an interlinked visual presentation of all meanings of a lexeme
combined with all the expressions it entails. An ICM possesses several prototypes
with decreasing importance, so, one prototype includes another, which, in turn,
includes another, and so on. Andreas Blank gives the example of the German
category VOGEL which is depicted in the appendix of this paper12. The ICM is
very helpful for including all associations of a lexeme and for explaining
metaphorical or metonymical usage of a word. With his model, Lakoff also
emphasizes the idea of ‘prototypical prototypicality’ as surmised by Geeraerts.
_________________12 Please look at figure 6 of the appendix.
3 Conclusion 15
3 Conclusion
The Polish linguist Anna Wierzbicka points out that “prototypes save”
(Tsohatzidis 1990: 347). Many semanticists deploy the concept of prototypes to
account for the unpredictability or lack of defining lexemes. In her opinion, the
notion of the prototype has been abused in order to replace complex explanations
and analyses. She does have a point in the sense that the prototype theory is
notable for its vague nature. Many problems of categorization are still not solved
by the theory. However, the notion gives considerable insight into the
unfathomable discipline of human cognition and offers an explanation of semantic
reference. Despite its shortcomings, the prototype theory has been very useful for
semantics, foremost by refuting the classical approach. It has proved that
categories can not be defined so precisely so as to draw a clear distinction
between them. With help of the extension of the prototype theory, some difficult
aspects, such as the inclusion of metaphorical usage, can be solved – even though
these theories are not in themselves fully satisfactory, as well. The prototype
theory should be considered as a key which has opened up many doors to a wide,
mainly undiscovered field of semantic categorization. Much research must yet be
done to explain the complexity of mankind’s perception and its reflection in
language.
4 References 16
4 References
Aitchison, Jean. 1987. Words in the Mind: An Introduction to the Mental Lexicon.Oxford: Blackwell.
Best, John B. 1999. Cognitive Psychology. (5th ed.). Belmont: Wadsworth. pp. 390-404.Blank, Andreas. 2001. Einführung in die lexikalische Semantik: für Romanisten.
Tübingen: Niemeyer.Cruse, D. Alan. 2000. Meaning in Language: An Introduction to Semantics and
Pragmatics. Oxford: OUP.Eysenck, Micheal W./ Keane, Mark T. 2000. Cognitive Psychology: A Student’s
Handbook. (4th ed.). Hove: Psychology Press. pp. 279-291.Kleiber, Georges. 1993. Prototypensemantik: Eine Einführung. Tübingen: Narr.Löbner, Sebastian. 2002. Understanding Semantics. London: Arnold.Mangasser-Wahl, Martina. (ed.). 2000. Prototypentheorie in der Linguistik:
Anwendungsbeispiele – Methodenreflexion – Perspektiven. Tübingen:Stauffenberg.
Persson, Gunnar. 1990. Meanings, Models and Metaphors: A Study in Lexical Semanticsin English. Stockholm: Almqvist & Wiksell International.
Rosch, Eleanor/ Lloyd, Barbara B. (eds.). 1978. Cognition and Categorization. Hillsdale,NJ: Lawrence Erlbaum Associates.
Schmid, Hans-Jörg. 1998. “Zum kognitiven Kern der Prototypentheorie.” In Ungerer,Friedrich. (ed.). Rostocker Beiträge zur Sprachwissenschaft, Volume 5:Kognitive Lexikologie und Syntax. Rostock: Universität Rostock. pp. 9-28.
Taylor, John R. 1995. Linguistic Categorization: Prototypes in Linguistic Theory (2nd
ed.). Oxford: Clarendon Press.Tsohatzidis, Savas L. (ed.). 1990. Meanings and Prototypes: Studies in Linguistic
Categorization. London: Routledge.Ungerer, Friedrich / Schmid, Hans-Jörg. 1996. An Introduction to Cognitive Linguistics.
London: Longman.Wittgenstein, Ludwig. 1963. Philosophical Investigations. Trans. by G.E.M. Anscombe.
Oxford: Blackwell. pp. 31e-32e.Zitzen, Michaela. “On the Efficiency of Prototype Theoretical Semantics.” English
Faculty, Heinrich-Heine-University, Düsseldorf, Germany. URL:http://www.ang3-11.phil-fak.uni-duesseldorf.de/~ang3/LANA/Zitzen.html (23July, 2002)
5 Appendix 17
5 Appendix
Cover picture: Aitchison 1987: 52.
Figure 1: Examples from Rosch’s GOE-rating tests (1975). Ungerer/Schmid
1996: 13.
5 Appendix 18
Figure 2: Birdiness Rankings, Aitchison 1987: 54.
5 Appendix 19
Figure 3: Attributes of the members of the category BIRD, Taylor 1995: 24
Figure 4: Selected common attributes and family resemblances of the
category BIRD, Taylor 1995: 27.
5 Appendix 20
Figure 5: Prototypes of verbal categories, Tsohatzidis 1990: 373.
Figure 6: ICM of the German category VOGEL, Blank 2001: 51.