Form, meaning, and reference in natural language: A phenomenological account of proper names

In my book Eigenname und Bedeutung (1996), I started from the observation that modern theories of proper names fail to do justice to the specific and complex semantic nature of proper names. Since the 1960’s and 1970’s, theorizing about proper names has been dominated largely by scholars working in the tra-ditions of analytic philosophy and logic, in particular John R. Searle and Saul Kripke. I argued, however, that the highly specific kind of meaning typical of proper names should be studied within a theory more in touch with general linguistics proper. The main philosophical (especially referential) and logical (especially formal) accounts start from the assumption that a proper name is “backed up” by encyclopaedic information held by speakers of the referents (Searle), or that a proper name is a meaningless, yet rigidly designating sign (Kripke). In contrast to these views, I argue that a general linguistic definition of the proper name has to focus not only on logical and philosophical issues, but also on the specifically linguistic semantic function of the proper name as a “part of speech” in actual utterances . This approach has nothing to do with pragmatics or discourse analysis, but aims at describing proper names and appellative nouns as categories of speech in language use, bringing into play a functional focus on proper names that has largely been lacking in definitions of the proper name so far. An outline of a semantic theory of proper names is then proposed based on some aspects of a “phenomenology of language and linguistics” as found in the work of Edmund Husserl and Eugenio Coseriu. Roughly speaking, Husserl represents the general epistemological implications of the paper and Coseriu its specifically linguistic aspects.


Introduction: Epistemology, phenomenology, and linguistics
Many scholars now working in linguistics, philosophy, and logic seem to agree that an adequate theory of proper names is an experimentum crucis for any comprehensive theory of language. Any theory of language unable to account for the properties of proper names and their function in language use must be considered of limited value. Moreover, scholars have come to realize over the past three to four decades that the nomen proprium is more than just a fascinating language universal. Proper names are complex signs with specific linguistic, pragmatic, logical, philosophical, semiotic, historical, psychological, social, and juridical properties, and hence represent a vast interdisciplinary field of study. On the other hand, the complexity of proper names seems occasionally to have yielded premature and one-sided definitions or ill-founded explanations that focus on one or two of the properties just mentioned but turn out to be incompatible with more general assertions about proper names. A truly integrative, comprehensive theory must therefore not only be broad enough to match the complexity of its subject matter, but also based on a well-founded general theory. In the present paper, I argue in favour of a theoretical framework that takes ordinary language use as its starting point (cf. Coseriu, 1955bCoseriu, , 1985Coseriu, , 1988, everyday speech and utterances in all their complexity being the original level of linguistic activity in which proper names occur and are to be studied.
Every linguistic phenomenon can be approached from different directions, and this is no less true of proper names. Moreover, proper names can be studied not only from various linguistic points of view (syntax, semantics, pragmatics, psycholinguistics, neurolinguistics, etc.) but non-linguistic ones as well, be they logical (as in Searle, 1969;Kripke, 1980;Devitt, 1976, and many others), philosophical (Gardiner, 1954), psychological (Landgrebe, 1934), stylistic (Aschenberg, 1991), etc. This diversity of approaches belongs in turn to the subject matters of yet another discipline, the meta-discipline called "epistemology". Generally speaking, the epistemology of linguistics is concerned with the relation between language as an object of study, on the one hand, and the scientific approaches, methods, and aims, on the other. Although I cannot enter into a historical survey here, it must be pointed out that different kinds of epistemology have developed in linguistics over the past one and a half centuries. One general theory in philosophy, logic, and psychology that in the course of the 20 th century has adapted to the tasks and challenges of linguistic epistemology, is phenomenology. Rather than give a detailed overview of the phenomenological approach to the epistemology of linguistics (see Willems, 1994 and, I will draw attention to those three aspects of it that are of special interest for the purpose of the present paper. First of all, phenomenology aims at determining, as precisely as possible, the part played in linguistic investigations by our preliminary linguistic intuition. The basic idea is that, like all language use, linguistics cannot but rely on what we already know about language intuitively (Coseriu, 1958;Itkonen, 1983). From a phenomenological point of view the challenge of a science like linguistics consists in converting this intuitive knowledge into explicit, scientific knowledge that is methodologically sound and well-founded without, however, running counter to just that intuition. Secondly, linguistic phenomenology regards the problem of meaning as central to both linguistics and the philosophy of language 1 , and recent developments in both disciplines seem to justify this view. To the phenomenologist, meaning is the central part of speaker intentionality (Husserl, 1929(Husserl, , 1939(Husserl, /1985. In other words, the focus on meaning is a focus on the central "condition", or resource, of linguistic activity, i.e. the origin of speech (in a non-genetic sense), as Husserl would have put it. Thirdly, for phenomenology an epistemological critique is never an end in itself, be it philosophical, logical, or otherwise. Instead, its ultimate purpose is to gain deeper insights into the object of the particular scientific inquiry at issue, which in the present case is linguistics. The ultimate purpose of a phenomenology of linguistics is thus to clarify the object itself, not (or at least not in the first instance) the theory and meta-theory of research, although in order to reach that end, the phenomenological focus must consider all relevant levels of inquiry and the role they play in elucidating scientific problems and objects.
In this paper, I address a specific problem in modern theorizing about proper names, and in doing so, hope to contribute to a better understanding of the semantic nature of the linguistic phenomenon we traditionally call "proper name". I start from the assumption that such a clarification forms a necessary prerequisite of any comprehensive theory of proper names as mentioned before (for more discussion, see Willems, 1996), indeed of any linguistic theory of proper names, and hope furthermore to demonstrate the wider significance of the issue to the theory of linguistic meaning in general.

The issue
The past few decades have witnessed several serious attempts at formal definitions of the proper name in logic and analytic philosophy, and at pragmatic definitions in (applied) linguistics. These approaches, the one more speculative, the other basically empirical, have succeeded at complementing each other in a great number of books and articles 2 . This is not surprising, since from an epistemological point of view the two approaches seem to share at least one basic assumption. Ever since the early 1970's, when Saul L. Kripke defined the functioning of proper names as "rigid designation" based on a "baptismal act" and ensuing language use (see Kripke, 1980), it has become almost commonplace in the theory of proper names as well as in empirical studies in onomastics to conceive of proper names as special semiotic devices with particular pragmatic purposes, used for referring to entities in the extra-linguistic world. With Kripke's approach, which became widely known, the problem of meaning was emphatically relegated to the sidelines of proper name theory. Kripke does not want to explain the meaning of a proper name in relation to its linguistic form. Rather, he is interested in the content of a referential and, ultimately, sociolinguistic tradition in the use of a proper name in a certain community of speakers. Despite notable differences otherwise (after all, Kripke developed his approach in explicit opposition to Searle's "descriptive" theory of proper names), Kripke's theory is on this point remarkably similar to that of John R. Searle (1969) 3 . Neither Kripke nor Searle focused on the problem of meaning in proper names: like Kripke, Searle was concerned primarily with the question of how the informational unity and integrity of a proper name could be explained in a logically as well as historically valid sense. Indeed, what I refer to as "informational unity and integrity" must be distinguished from semantics proper. By asserting that a proper name corresponds to a certain set of (definite) descriptions that hold of the referent named, or, to put it more accurately, to a disjunctive class of such descriptions, Searle (1969, p. 167) reduced the semantic problem of proper names to a question concerning the informational status of proper names within knowledge in general, and language use. Although Kripke proposed essentially the same view some years later, he proceeded in an entirely different way, stating that proper names are non-descriptional and designate by "direct reference" 4 . Contrary to what has often been maintained, the contrasts between Kripke's "causal (chain) theory" and Searle's "cluster theory" do not arise from different views of the meaning of proper names. Rather, they relate to the way in which both scholars define the informational content that is, in one way or another, assigned to proper names as particular devices in language use-and that is a different matter altogether.
In the theory of proper names both the Kripkean and Searlean approaches are indebted to a logical as well as referential mainstream in modern philosophy of language (cf., e.g., Devitt & Hanley, 2006). The decisive impact of this mainstream in overall present-day semantic theory has not only revealed the virtues of analytic precision. This is particularly obvious in the theory of proper names. Today, a theoretical basis for an adequate semantic analysis of the proper name as a part of natural language is still missing, which is particularly obvious when compared with other areas in semantics and syntax. Such a basis, however, has to complete the discussion in matters that are not being dealt with in the debate between (adherents of ) S. Kripke and J. Searle. In the present paper, I argue that a theory of proper names is incomplete unless it manages to account for the genuine linguistic relation between linguistic form and linguistic meaning, as defined, e.g., by Saussure (1916Saussure ( /1968 in his influential, though not always properly understood, outline of a theory of the linguistic sign. My own account of this relation will be founded on the theories of proper names outlined by Edmund Husserl (1859-1938 and Eugenio Coseriu (1921Coseriu ( -2002. While unfortunately much less well known, each in his own way considers proper names to be a particular kind of complex linguistic sign in which both form and meaning play an essential role. I will focus on Husserl as an epistemologist whose phenomenology is, among other things, an impressive attempt at an illuminating 2 For bibliographical details, see Eichler et al. (1995Eichler et al. ( -1996. 3 I will not enter into historical details here. Suffice it to say that I contrast, for argument's sake and in a somewhat simplifying and generalizing manner, the tradition in which Searle is to be situated (G. Frege, B. Russell, L. Linsky, among others) with the tradition to which Kripke belongs (along with K. Donnellan, D. Kaplan, H. Putnam, among others); cf. Salmon (1996Salmon ( , p. 1142. 4 Not surprisingly, in Chapter IX of his book Intentionality (1983), Searle tries to convince the reader that Kripke's causal theory too is founded, though implicitly, on a descriptive argument; see also the discussion in Dummett (1996) and Salmon (1996, §6). The differences between the notions of rigidity as used by S. Kripke and D. Kaplan are explored in Steinman (1985). dialogue between philosophy and other sciences such as physics, logic, and linguistics. Coseriu will be introduced as a major linguist and theoretician of language of the 20 th century who has made one of the most intriguing contributions to the interdisciplinary discussion between philosophy and linguistics with regard to the theory of proper names.
3. The phenomenological difference between "mere form" and "linguistic form" The fact that neither Kripke nor Searle discusses the relation between form and meaning proper serves as an emphatic reminder of the sharp distinction that must be made between "form" (without further specifications) and "linguistic form". Note that the term "form" as such is unspecific: one can speak of "forms" in biology as easily as in logic, art or physics. The term "linguistic form", by contrast, is highly specific. It refers to the natural languages of human beings, to the extent that we do not only have to distinguish several phonological, morphological, lexical, and syntactic forms within one language (the "linguistic forms" of phonemes, morphemes, words, and sentences), but also the "inner form" of one language as opposed to another language (in the tradition of Humboldt, 1830Humboldt, -1835Humboldt, /1963. As a consequence, "linguistic form" is always and necessarily the "form" of a particular historical language; linguistic universals and types, too, must be realised in particular historical languages, e.g. English, Chinese, or Arabic. Obvious as this observation may seem, in the theory of proper names the distinction between "form" and "linguistic form" is an indispensable prerequisite. First of all, proper names belong to natural language. This means that speakers in different linguistic communities know what proper names are on the basis of their natural linguistic knowledge (intuitive competence), something that can be ascertained simply by observing speakers generating proper names in everyday language use and identifying them as proper names whenever they encounter them. From a phenomenological point of view, the famous "singular terms" of logicians (G. Frege, B. Russell, A. Tarski, D. Davidson, among others), which are often presented as the most perfect examples of proper names (for example the deictic l or the definite description the capital of France), are no proper names at all. Instead, their definition as "singular terms" derives from proper names already existing in natural language, not vice versa, due e.g. to some striking semantic and/or referential as well as pragmatic similarities between proper names and the logicians' "singular terms". In particular, we cannot reasonably claim that proper names in ordinary language are a sort of "deficient" realisations of some ideal and essentially non-linguistic device that is the object of, for example, logic (cf. Frege, 1892). The reverse is the case: the "singular names" referred to by most logicians (definite descriptions, indexicals, etc.) are really just radically unambiguous forms, derived on the basis of an interpretation of the proper names that already exist in natural language, and of the primary linguistic competence corresponding to such primary names in historical linguistic communities and their languages.
Secondly, and more importantly still for the purposes of the present paper, the distinction between "form" and "linguistic form" is a mere corollary of the fact that natural, ordinary language is more than just a formal phenomenon, linguistic forms always being, in one way or another, meaningful. As I pointed out earlier, this view is at the centre of linguistic phenomenology. Precisely because form is, metaphorically speaking, only one side of the linguistic sign, the reverse being its meaning, the theory of proper names does not deal with mere "forms" of signs but with full-fledged "linguistic forms", i.e. form-cum-meaning. Admittedly simplified as this epistemological reasoning is, it shows that any theory of proper names as meaningless signs is a priori untenable.
In order to illustrate the impact of the distinction between "form" and "linguistic form" on theorizing about the meaning of the proper name, it is particularly revealing to take a closer look at the analysis of the process of naming in Kripke's causal theory, and to describe in more detail in what sense Kripke deals with "forms" instead of "linguistic forms". In Kripke's view (adopted in a great number of recent studies) the reference of a proper name is fixed by a "baptismal act" of designation. The proper name is thus conceived of as a form defined in terms of its informational status and corresponding truth value. For instance, the form "John" is attributed to a person X, and this particular (and verifiable) information is then passed on (via a "causal chain") to other members of the linguistic community both synchronically and diachronically. Kripke's theory is thus essentially based on form and reference combined in the act of naming. The act of naming is conceived of as a kind of linguistic activity, relating a referent to a form, yet it is not necessary for this form to be expressed in natural language (be it in speech or writing). Some other sign-a mark, a card, or any other symbol-could do the job just as well, if only the labelling and the subsequent transfer of information are successful and adequate. In short, in Kripke's theory the naming is done in an immediate way, to the extent that there is no meaning-function attributed to the form of the name other than the referent 5 . Using a helpful distinction introduced into linguistic theory by Coseriu (1955a), one can say that in Kripke's theory no "full form" is linked to a referent but only an abstraction of such a full form, which I call a "mere form". On the basis of the principle that the genuine linguistic sign in natural language is a bilateral sign, consisting of a form that corresponds to a meaning, Kripke's form is "just" form, form without meaning, despite being causally linked to a referent. What corresponds to the "form" of the name, by definition or by convention ("fiat", cf. Pendlebury, 1990, p. 522), is not meaning but some referential information (i.e. information based in an act of referring), and this information is the kind of "content" that induces the chain of rigid designation. Kripke's theory clearly leaves no room for intra-linguistic meaning, the relevant "semantic" aspects of the act of naming being reduced to nonlinguistic, referential information. In this respect, Kripke's theory is very similar indeed to Searle's, even though Kripke explicitly rejects the encyclopedic "descriptive backing" proposed by Searle as a basis for proper names.
Putting it mildly, Kripke's theory must be considered one-sided and incomplete (though not altogether misconceived, see §5.1) from the phenomenological point of view, since it does not account for the relation between proper names and the essential character of ordinary language in general, viz. that form and meaning always mutually presuppose each other in the linguistic sign. Therefore, Kripke's theory cannot be said to deal with proper names in natural language at all, and Ziff (1977, p. 328) is right to state: "It is only in a formal language indeed a modal logic that one could hope to find anything that qualifies as a rigid designator". This brings us finally to an even more fundamental issue that is of considerable importance for linguistic theory in general. Eventually, Kripke's account seems to imply that proper names, although parts of everyday natural language use, are in one way or another exceptions to the form/meaning relation characteristic of natural language in that they dispense with the sign's meaning component altogether. This amounts to claiming that proper names are no linguistic signs at all but rather general signs with a particular pragmatic and referential function, yet without any intra-linguistic semantic function. In §5.2 I will return to this issue.

Different kinds of meaning in natural language
Kripke's theory (just like Searle's) was not conceived in a historical vacuum. Ever since John Stuart Mill have philosophers, logicians, theorists of language, linguists, and onomasticians claimed that proper names have essentially no meaning but merely reference. As shown in the previous section, this view implies that proper names are conceived of as "mere forms", not meaningful "linguistic forms", and that proper names function as labels marking extra-linguistic entities. The view that proper names are essentially meaningless forms has gained widespread acceptance not only in the theory of proper names but also in the theory of language in general ever since the 19 th century 6 , yet to the phenomenologist, this very fact indicates a semantic problem that is as deep-rooted as it is misunderstood.
If the semantic dimension of the problem of proper names has largely remained unresolved, this is undoubtedly in large part due to the fact that the term "meaning" that crops up time and again in discussions is so ill-defined and vague. As a consequence, proper names could even be denied meaning altogether as a result of Jakobson's observation (1957/1971, p. 131, based on Bertrand Russell) that, for instance, all dogs called Fido do not share any property 'Fidoness' , whereas words like pup, mongrel, hound do have a general meaning that can be indicated by the abstract nouns 'puppihood' , 'mongrelness' , and 'houndness' , respectively. On this view, the word Fido does not semantically imply anything like 'belonging to the class of Fido's'; on the other hand, if the word dog is used to designate an individual dog or a species, then this use of dog indicates that the referent is conceived of as belonging to the class of canines. Furthermore, the word dog can be used to designate any specimen of the class of canines, whereas not every dog is called Fido.
What does Jakobson's observation tell us about proper names? Obviously, there is an important distinction to be made between a proper name and an appellative, based on the fact that the proper name lacks a class-meaning, i.e. a meaning that enables the speaker to subsume the referent under a certain class. In the present article I use the term classematic meaning to refer to this kind of meaning 7 . But does the assertion that proper names lack classematic meaning imply that they have no meaning at all? Clearly, such a conclusion would be premature because Jakobson only demonstrates that the meaning of proper names must be of a different kind compared to the meaning of appellatives. The key question, then, is how to define this difference in meaning.
It goes without saying that the distinction between several types of meaning is not only relevant but even indispensable, and this is particularly clear if we take a closer look at the semantics of proper names. Although many scholars (e.g. G. Leech, J. Lyons, K. Allan, among others) have attempted to differentiate between kinds of meaning, their accounts remain in a fairly traditional vein and cannot be said to bear upon the specific semantic nature of proper names. One of the few accounts of immediate relevance to the semantics of proper names is the functional theory of meaning outlined by E. Coseriu in several articles and books. It seems useful therefore to dwell upon what Coseriu has to say in this matter, all the more so because his account does not appear to have been fully understood in recent studies (cf. Willems, 1996 for discussion).
The first thing to be clear about from the outset is that the kind of meaning usually designated by the general term meaning is what Coseriu calls the lexical meaning of words or lexical morphemes, i.e. the kind of meaning distinguishing come from go, white from green, house from building, plant from tree, -less from -ful, and so on. Lexical meaning can be said to be "classematic" in the sense introduced above. For example, as an element in the lexicon of the English language the word house does not mean a particular house, a group of houses, or several types of existing houses, etc. The meaning (Bedeutung) of the word is not its reference (Bezeichnung). Just like its form, the lexical meaning of the word house is part of the English language, and of this language alone. In epistemological terms, the meaning of a word can be called the condition of its use, for two reasons. First of all, the meaning of a word is an ideational, intersubjective "rule" that a speaker has to apply if she or he intends to refer to an object in the external world that falls under that meaning. In natural language there can be no reference without meaning as a basis of linguistic activity, which is both speaking about something and speaking with someone (Coseriu, 1994, p. 82-83). Secondly, the meaning of a word is a general and a historical concept at the same time. Meaning is, in this sense, a language-specific resource, for instance the concept a speaker of the English language has in mind when he produces the form "house". The concept corresponding to this form is a virtual "class", the class thinker makes use of them that they stand for anything, or, in one sense, have 'meaning'" (Ogden & Richards, 1945, p. 9-10). It is this confusion between 'meaning' and 'standing for something' that has plagued both the theory of language and the theory of proper names for a long time. 7 I deliberately avoid to use Bloomfield's (1935, p. 146, 202) term class-meaning here, as Bloomfield-not surprisingly, within his behaviorist framework-does not distinguish clearly between the lexical level of the historical langue and the categorial level of discourse in general (see also below for the notion classematic meaning).
As Coseriu (1987, p. 149) rightly points out, however, the lexical meaning is but one type of linguistic meaning, even though it is a very important one. All in all, Coseriu distinguishes five different types of meaning that are all the semantic counterparts of linguistic forms-forms that, as we will see below, can be able to convey more than one semantic function (which has nothing to do, however, with polysemy as is commonly understood, see Coseriu, 2000).
Apart from the lexical meaning just mentioned, a second type of meaning-a particular important one in view of the purpose of the present paper-is the meaning that is common to, e.g., the adjectives white, green, red, and that differs from the meaning common to the corresponding nouns whiteness, greenness, and redness. Of course, there is a difference in lexical meaning between e.g. white and green, or between greenness and redness. Yet, the difference in meaning between the series of adjectives and the series of nouns does not lie in the lexicon, even though the items registered in the lexicon may partly reflect this difference. However, the semantic difference at issue arises from the difference between the parts of speech we call "adjective" and "noun" (or "substantive"). This difference is based on the intentional modes in which the subject matter of speech, "reality" in its widest sense, is represented. Parts of speech are semantic categories based in rational linguistic activity as such; they are, for that matter, potentially universal, whereas lexical meaning is necessarily language-specific (though possibly cross-linguistically general). Consequently, in phrases like a white wine the word white carries at least two different types of meaning in ordinary language use, viz. a lexical meaning because it is an item of the English lexicon, and the value of being an "adjective". In this sense, the form of the word can be said to correspond to more than one meaning function. In accordance with traditional epistemological terminology, Coseriu calls the kind of meaning that corresponds to a particular part of speech categorial meaning. From the phenomenological point of view, it is important not to confuse a category with a class (see Cobb-Stevens, 1990, p. 148-154, 169), or, in other words, to confuse categorial meaning with classematic meaning. One may, for example, attribute a lexical class-meaning to the word fire, yet the categorial meaning of this word only shows in actual speech, or abstract representations thereof. Whether fire is used as a verb (for example in the sentence They fired rubber bullets) or as a substantive (for example in The troops came under fire from both flanks) depends on the sentence as a whole (see also Coseriu, 1987, p. 27, and2004) 8 . I return to the definition of the categorial meaning of proper names in §5.
A third type of meaning is called instrumental meaning. With this term, Coseriu not only refers to the semantics of combinatorial procedures and elements such as word order, intonation, affixes and desinences, but also to the meaning of typical "function words" such as articles, prepositions, and conjunctions. If one says the book, the lexical unit book is said to be "actualized" by means of the definite article, in a way that is different from "actualization" by means of, e.g., the indefinite article (a book); it is also different from "pluralization" by means of a bound morpheme (book-s), or from specifying "delimitation" realized, for example, by means of a preposition (e.g. without books) 9 .
The fourth type of meaning is called syntactic meaning by Coseriu. Syntactic meaning combines lexical and/or categorial meaning with instrumental meaning. For example, modes and tenses of verbs (e.g. indicative as opposed to imperative, or the present indicative as opposed to the future indicative), voice (active and passive or antipassive), etc. correspond to differences in syntactic meaning. The claim is therefore erroneous that there is no real difference in meaning between the following two sentences: (1) Hamlet killed Polonius.
Although it is legitimate to maintain that there is no difference in lexical meaning between both sentences nor in reference, their overall meaning is not identical, since their different syntactic forms result, e.g., in different topic-focus marking.
Finally, the fifth type of meaning that, according to Coseriu, should be distinguished within the general concept of meaning is ontological (or "ontic") meaning. With this term, Coseriu refers to differences in the various representational values that derive from differences in syntactic constructions, for example the different values expressed in (3) and (4): (3) How tall! as opposed to: (4) How tall?
These sentences share the same syntactic meaning but the "existential value" attributed to the state of affairs expressed in both sentences is different. Thus, various ontological meanings correspond to differences like those between an affirmative sentence and a negation, an exclamatory sentence and an interrogative sentence, etc. These differences form the bridge from the level of the language system (langue) to the level of text linguistics, given that it is important to determine whether or not there are particular languagespecific "meanings" that correspond to the different "sense functions" that express ontological meanings 10 .
The distinction between five types of meanings presented above implies that differences in linguistic form never are merely formal. Formal differences always entail differences in meaning: not necessarily differences on the lexical level, to be sure, but always on at least one of the five levels just described. Furthermore, such semantic differences cannot be reduced to reference, and it does not come as a surprise that we also find differences in meaning with respect to the "classematic" nouns of the lexicon as opposed to "non-classematic" proper names, both in actual speech and with respect to the linguistic knowledge speakers possess of their language. The latter point is particularly important. A clear distinction must be made between what I call "actual speech" and the kind of knowledge human beings have of one or more historical language systems. As I will argue in later sections of this paper, proper names, i.e. proprial parts of speech with a categorial meaning, cannot be listed "taxonomically": only "forms" of proper names can. Taxonomies are of course common practice with respect to lexicalized appellatives in any lexicon, but as I noted above, appellatives are morphemes with classematic meanings. Since proper names lack such meanings, it is doubtful whether a list of proper name forms can tell us anything about the true semantic nature of proper names as parts of speech.
Before pursuing in detail the semantic definition of the proper name in the next section ( §5), I would like to summarize the line of reasoning so far. Firstly, the difference between "classematic" and "nonclassematic" words is not to be found on the level of lexical meaning. Instead, this difference relates to categorial meaning, i.e. the level of the parts of speech. This level refers to universal modes of "moulding" reality in language, modes that correspond to meaningful (indeed "full") linguistic forms. Secondly, in elaborating a semantic theory of proper names, much more is involved than just differences of a referential and pragmatic nature. This is particularly important if one intends, as I do in the present paper, to explain the conditions of reference and language use by means of proper names, in accordance with the phenomenological approach outlined above.

Meaning and referring according to Edmund Husserl
In the previous section I explained why the difference in meaning between e.g. dog and Fido is not a difference in lexical meaning, although both are nouns. For a full account of the difference between both nouns, one has to consider in detail the specific status of what I call, following E. Coseriu, the categorial meaning of a proper name.
In the sentence The dog was barking at least two types of meaning are united in the word dog, viz. the lexical information of the English word ('domesticated carnivorous mammal etc.') and the categorial information "noun" the word comprises by virtue of being a particular part of speech in the sentence. In Fido was barking the word Fido again clearly displays the categorial status of a noun, yet it lacks the lexical meaning found in the NP (the) dog in the first sentence. In particular, one should not commit the mistake to consider the information Fido is a dog as part of the categorial meaning of Fido. First of all, in considering the sentence Fido was barking, one must be aware not to project the kind of lexical information rendered by the predicate was barking into the NP Fido, thus obtaining the classematic specification "Fido is a dog" (see note 8). Secondly, the categorial meaning of Fido, on the one hand, and what as speakers of the English language we know about creatures commonly called "Fido", on the other, not only are two entirely different things, but the latter does not bear on the former. Although the categorially specified word Fido is lexically empty as a proper name, it is very well possible that most speakers of English consider Fido to be a word mainly used as a name for dogs. However, this is no lexical meaning that could be assigned to Fido. Rather, it is information speakers derive from ordinary language use in a particular historical setting in which Fido regularly occurs. This is precisely why it is so important to distinguish the lexical meaning of a word from the pragmatic knowledge corresponding to normal language use in some setting.
Leaving aside the problem of etymological opacity, the most conspicuous characteristic of the fact that the proper name Fido lacks lexical meaning, is that there are no paradigmatic linguistic relations between Fido and other proper names. Categorially, Fido is neutral in respect to what living being, what thing, or what state of affairs is called "Fido". Although it is true that Fido is usually the name of a dog, it could also be the name of a cat, a tornado, a disaster, or something else altogether. There are no intralinguistic categorial restrictions concerning this issue of the referential domain, precisely because Fido does not fill a slot in a lexical paradigm. The lexeme dog, on the other hand, does just that. The word is part of a larger field of lexical units, a lexical field or lexical paradigm (cf. Coseriu, 1978, p. 195) 11 . For example, the word dog is subordinated to mammal, animal, creature, etc. Furthermore, the meaning of dog is related to and distinct from the meanings of words like horse, cat, etc., while it can be intrinsically (i.e. classematically) determined by further specifications such as tall, white, angry, etc. All this does not hold of Fido, as there is no purely linguistic reason that Fido ought to be a mammal, a dog or a creature with a certain size, appearance, emotional state, etc. As a consequence, the relation between words like dog, horse, cat is entirely different from the relation between, e.g., Fido, Blacky, and Benjy. The difference is that the words in the first series are primarily related on intra-linguistic grounds, whereas the second series contains items primarily related on the basis of extra-linguistic knowledge, which of course can become (partly) conventionalized in normal language use (cf. Coseriu, 1952). It is unquestionably a great merit of S. Kripke's causal chain theory to have paid due attention to this particular difference, yet in Kripke's theory the linguistic aspects of the difference remain obscure.
Above all, with regard to linguistic content, a "paradigm" should not be confused with a "series". If a linguistic item belongs to a "paradigm", this means that the item is a unity of a form and a meaning (for example a lexical, morphological, or syntactic item) that occupies a place within a system or subsystem of a particular historical language. As a consequence, filling a "slot" in a paradigm always entails an opposition on some categorial level of language: man, woman, boy, girl; hot, warm, cold, tepid; look, see, watch, perceive; etc. By contrast, there is no categorial restriction as to a linguistic "series", for example the series of typically feminine given names ending in -a in several Germanic languages (Alberta, Joanna, Amanda, Patricia, Claudia), or series of place-names (Oxford, Chelmsford, Hertford, Watford, Guildford, Stratford), etc., for in sentences such as I love another Amanda and I have not seen your Oxford neither! the words Amanda and Oxford are no proper names (see §5.3 and §5.4 for further discussion).
If it is correct that the proper name Fido has no lexical meaning in the traditional sense of this term, then it is important to realize that there can also be no difference in lexical meaning between dog and Fido. The importance of this distinction has been expounded with great talent by Edmund Husserl, the founder of modern phenomenology. In his Logical Investigations, Husserl argues that the relation between "meaning" (or "signification") and "referring" is radically different if one compares proper names to appellative nouns (Husserl, 1900(Husserl, -1901(Husserl, /1984, II, 1, I, §12) 12 . According to Husserl, the reference of words like horse is very broad because such words are what he calls "universal names". This means that we can use them to designate ("refer to") an indefinite number of referents in the external world. As a result, horse is a word with a vast "extension" ("vielumfangend" to use Husserl's own term). The case is different for proper names such as Bucephalus. Words like these are no "universal names", their use as proper names implies that they do not designate classes. Consequently, Bucephalus cannot, as a proper name, be "vielumfangend", but only "vieldeutig", equivocal or ambiguous, for example if other horses are called "Bucephalus" besides the one belonging to Alexander the Great. In terms of phenomenology, the possibility of equivocalness exists as far as the "mere form" of the proper name is concerned, i.e. the formal abstraction from the proper name as a "full form". In other words, as a "full form" in actual speech (i.e. as a linguistic form bearing meaning) a proper name is never ambiguous (barring possible misunderstandings or intentional "double entendres" in dialogue, on discourse level, that do not matter here) (Husserl, 1900(Husserl, -1901(Husserl, /1984, II, 1, IV, §2). "Fido", for example, can be an ambiguous form in abstracto because more than one referent can have this name. Yet this does not mean that one would be unable to assign a particular referent to Fido each time the word is intended as a proper name in discourse, nor would it entail that all the referents called "Fido" have a linguistic feature 'Fidoness' in common. Thus, the equivocalness of "Fido" does not relate to its being a proper name but to the fact that one can always reduce a "full" proper name to its mere form ("F-i-d-o") and subsequently assign a classematic meaning to it, as in I know another Fido (e.g. with the meaning 'I know another dog called "Fido"'). By contrast, all creatures that are subsumable under the word dog share the common feature of being a canine simply because of the lexical, classematic meaning of the word dog 13 . Unambiguous forms of proper names corresponding to only one referent are, of course, conceivable: let us suppose for the sake of argument that this is the case with Popocatepetl (the volcano in Mexico) or Tchomolungma (the Tibetan name of Mount Everest). But this does not obscure the essential semantic difference between proper names and appellative nouns, this difference having, as Husserl knew, nothing 12 Rosado Haddock (1982, p. 432), it is wrong to suppose, as D. Føllesdal and others do, that Husserl owes his distinction between sense and reference to Frege. I will not enter here in a discussion of these terms (as well as related terms like intension and extension) common in the writings of Frege, Carnap, Wittgenstein, Searle, Kripke, Kaplan, among others. My purpose, as previously stated, is to broaden theorizing about the meaning of proper names by focusing on a number of basic phenomenological ideas about the study of language and linguistics that have been neglected in current discussions of the theory of proper names.
13 This is also true if the word dog is used metaphorically (e.g. as a designation of a brutal person) because the metaphorical use of a word (e.g. a common noun), entails its paradigmatic meaning, it does not suppress or alter it (cf. Coseriu, 1987, p. 109).
to do with (potential) unequivocalness. Both proper names and appellative nouns are full-fledged words, "full" linguistic forms, though with different types of subcategorial meaning within the overall category of "noun". However, strictly speaking, only words like dog, horse, cat, etc. are "lexemes", if one takes the term "lexeme" to designate the class of fully linguistic morphemes in the lexicon of a particular historical language. On the other hand, Fido is no "lexeme", even though it has a form commonly used in one or more historical languages as a name for a dog. Still, it is not the "word" Fido that is ambiguous but an abstraction of the word, its (reduced) "mere form"; and this "mere form" can in turn be taken as the form of a full lexeme via reinterpretation, as in All the Fidos I know are dogs (see §5.4).
Thus, in the Husserlian framework, both "universal" words and proper names have meaning. At the time of his Logical Investigations and of an early article on the logic of signs (written in 1890 but not published until 1970 14 ), Husserl calls appellatives "indirect signs" and proper names "direct signs" (see Husserl, 1900-1901/1984, II, 1, IV, §3, and 1890/1970. For Husserl the vital point is that "sense" (i.e. linguistic meaning) and "reference" coincide in the proper name, whereas they do not (and cannot) coincide in the appellative with its classematic meaning 15 .
There can be no doubt that Husserl's remarks on the highly specific semanticity of proper names are as valuable as they are coherent, yet they remain rather general from a linguistic point of view. It was E. Coseriu who analysed in more detail the difference between lexemes and proper names (Coseriu, 1955a). According to the definition of a proper name as suggested by Coseriu, proper names are monovalent, individualizing, and one-dimensional, whereas appellative nouns are polyvalent, generic, and twodimensional. It is worthwhile to take a closer look at these features, because the distinctions introduced by Coseriu deepen Husserl's general analysis considerably on quite a number of scores.

Monovalence: The non-classematic function of proper names
Recall first of all that, from the phenomenological point of view, a word is the combination of a form and a meaning, and that both lexemes and proper names are words in the full sense of the term. Monovalent words designate referents without classifying them as members of a class. This kind of designation appears to be characteristic of proper names. By contrast, an appellative noun is polyvalent, because the word applies to all members of the class that corresponds to the word's general ("universal") meaning. Hence, monovalence is no doubt one of the most important aspects in the semantic definition of proper names and the definition of proprial meaning in general, but it must be understood adequately. From Husserl's explanation of the difference between universal names and non-universal names, quoted above, it follows that the monovalence of proper names is not a particular subtype of classematic designation. Instead, it is a highly specific negation of classematic designation, such that proper names and appellative nouns share a basic categorial meaning (both being "nouns"), while proper names lack the classematic function lexemes typically have as parts of speech. The negation of classematic meaning, then, must not to be misunderstood as the negation of meaning as such: after all, proper names are meaningful words. The existence of classematic meaning is, moreover, a "logical" prerequisite for its negation. It follows that proper names must be regarded as genuine linguistic signs, not as general but essentially non-linguistic signs, as Kripke (and other scholars too) assume (see §3).
Given the epistemological focus of the present paper, it is important to understand precisely why proprial meaning should be regarded as a kind of negation of classematic meaning (to be distinguished from denial, which is the form of a proposition). This is for two reasons. On the one hand, theorizing about proper names in the logical tradition appears to start from the highly questionable assumption that proper names are no genuine linguistic signs and fails to shed light on specific semantic nature of the proper name as a part of speech. Yet it is precisely this kind of theory that is most pervasive in current research into the communicative and pragmatic aspects of proper names and their use in natural language, and that alone is reason enough to reconsider its premise. On the other hand, the view advocated by some scholars in the 1980's (cf. e.g. Seiler, 1983) according to whom a proper name denotes a class which number of members equals one, appears to be no less misguided. There can be no class with only one member: membership in such a class would simply eliminate the very idea of class and this, at least as far as language is concerned, is inconceivable. The claim that a proper name designates a class with only one member arises from one of the most enduring misconceptions in the theory of proper names: the idea that meaning can only be defined in the classematic terms of the lexicon of a given language. Even the idea that proper names are meaningless signs which can accumulate a certain amount of "significance" in the course of their use (Sonderegger, 1987, p. 16), derives from the same too narrow semantic principle according to which linguistic meaning in general must be defined on the basis of the lexicon. This is not to deny that the kind of "significance" Sonderegger is referring to undoubtedly exists: it is easy to see that the "associations, impressions, and feelings" which Sonderegger says are connected with proper names, can be conventionally related to some referential, extra-linguistic information eventually captured in the classematic meanings of lexemes 16 . Yet this "significance" is not part of the proper name as a linguistic category, and in seeking a semantic definition of the proper name, it would be quite inappropriate to resort to any imaginary analogies between its "significance" in the sense of Sonderegger and the kind of semanticity found in appellatives (but not proper names).
There is still another, more important reason why proprial meaning should be regarded as a negation of classematic meaning. It is true that proper names constitute a genuine linguistic category, or more precisely: a subcategory of the pars orationis noun. Yet because they are monovalent, it is impossible to identify them with the "primary names" of a natural language. Monovalent terms necessarily belong to a superstructure of language, and this structure is conditioned by and founded upon plurivalent, universal terms. The opposite claim, according to which proprial terms "precede" universal ones, may occasionally have found proponents throughout the history of proper name theory (see Willems, 1996 for more details), but it is nonetheless epistemologically incoherent: it presupposes an initial conception of the "world" (where "world" stands for all possible referents, Eley, 1972Eley, /1985 reminiscent of an atlas in a scale of 1:1, something which cannot result in meanings but only in a restatement of objects. Yet meanings are no objects. If a proper name, e.g. John, is assigned to a referent, the latter has to be "categorized" already by means of a polyvalent term, for instance as a 'living being' , a 'human being' , a 'man' , a 'boy' , etc. These terms are, to be sure, not language universals (cf. Lakoff, 1987): they are concepts corresponding to appellative nouns with classematic lexical meanings in a particular historical language (or a set of related historical languages). The vital point is that for any referent to be named by means of a proper name, this kind of meaning must be available, yet the part of speech called "proper name" is only founded upon this kind of meaning, the classematic meaning not being transmitted to the proper name, which after all is a monovalent subcategory. While the classematic categorization of the world does underlie the proper name, it is rendered ineffective in the "proper naming" because a proper name does not classify the referent to which it is applied as a member of any class 17 .

Individualization: The difference between external unity and individual representation
This brings us to the second element in Coseriu's account, the observation that a proper name individualizes a referent. This observation has an important corollary. Apparently, the unity or one-ness, or even the real existence of the referent ("before" it is designated by means of a proper name), is of no importance as to whether a word qualifies as a proper name or as an appellative. This is of major significance for three reasons. First of all, as far as proprial meaning is concerned, the referent that is individualized by means of a proper name exists as a "function" in the universe of discourse. From a linguistic point of view, such an "existence" in the universe of discourse (a "Sein", to use the philosophically laden German term) is not to be mistaken as a "Dasein" or "Seiendes" in an extra-linguistic sense. There is no linguistic or semantic difference, therefore, between Scylla and Charybdis, on the one hand, and Messina, on the other, since the referential differences involved pertain once again to the knowledge of the "world" rather than (the knowledge of ) language. Secondly, it has no influence on the linguistic status of the word, whether there is a referential, representational or even natural unity that corresponds to the individuating function of the proper name. The Bahamas, the Greater Antilles, the Lesser Antilles, and the Pyrenees are as much proper names as Cuba, Hawaii, Mexico, and Germany. The individualizing function of proper names also covers NPs like the Grand Canyon, thus representing a group of entities as a unity on the categorial level of speech. Thirdly, real existence and one-ness of the extra-linguistic referent being no prerequisites for a word or NP to qualify as a proper name, there is no requirement whatsoever for the referent of a proper name to be unique in one way or another. Husserl rightly saw that a proper name is equivocal only as far as its "mere form" is concerned (e.g. John, Bill, Margaret, etc.) but that its formal equivocalness does not cause the word to lose its ability to function as an unequivocal proper name in actual speech. Although there are many persons called "John" (or even "John Miller"), the part of speech John always individualizes one single referent each time it functions as a proper name. On the other hand, many people consider words like sun or moon to stand for unique referents and therefore to be proper names, yet in many languages (such as the Indo-European languages) the sun and the moon are not designated by proper names. Unlike their referential designata (within a particular, naive conception of the world, that is), the words sun and moon are generic in nature and, as a consequence, no proper names but appellatives.
In order to gain full clarity about the individualizing function of a proper name as a nominal subcategory, let us consider words like Skat (or skat), mannerism, (the) waltz, Marxism, quantum mechanics, and February. According to Dummett (1996Dummett ( , p. 1188, it is not clear whether these words are proper names or not. Whether words are proper names or not is of course impossible to decide when these words are studied in isolation. Once restored to their ordinary functions as parts of speech in actual utterances, however, it becomes perfectly clear that the words mentioned by Dummett cannot possibly be proper names (examples are mine, K.W.): (5) My uncle always wants to play skat. (6) They could not appreciate the mannerism of the author. (7) Vivian has never danced a waltz. (8) Paul always tries to refute Marxism. (9) Future generations will prove that the basic hypotheses of quantum mechanics are false. (10) They plan to move in February.
The italicized words in the above sentences do not individualize the object referred to without simultaneously classifying it. In (5) skat simply designates a particular game among other similar games, in (6) mannerism refers to some author's particular style of writing, in (7) (the or a) waltz designates a particular dance (of which there are various forms) accompanied by a certain genre of music. None of these words individualizes an object, an activity, a game, a style, etc. in a non-classematic way; they are all polyvalent words and hence appellatives. Furthermore, in (5), (6), and (7) all examples derive their meaning from being a part of a lexical "paradigm" to which they belong : skat, bridge, whist, twenty-one, etc.; mannerism, classicism, eclecticism, orientalism, etc.; waltz, foxtrot, Charleston, rumba, etc. In (8) and (9) Marxism and quantum mechanics are no proper names either, yet for a different reason. Marxism and quantum mechanics cannot be regarded as appellatives in the sense that skat, mannerism, and waltz are. Unlike the latter, Marxism and quantum mechanics primarily belong to particular nomenclatures, and being technical terms, they are distinguished from the ordinary words of natural language by their semantic definition, which corresponds to a piece of scientific knowledge rather than an intuitive "universal" linguistic concept that is a part of a lexical field. Yet Marxism and quantum mechanics are no proper names either. On the one hand, unlike scientific terms (like quantum mechanics) or technical terms (like H 2 O,DNA,AIDS,etc.), proper names form a category of natural language. On the other hand, when using words like Marxism and quantum mechanics, one does not refer to individuated instances, Marxism being a body of doctrine, and quantum mechanics being a theory just like Analytic Philosophy, Phenomenology, or Behaviorism.
Finally, while not a technical term, the word February in sentence (10) is no proper name either. One reason is that the word is obviously polyvalent, as it subsumes the referent into the class of months. The use of February does not correspond to an individualizing conception of a referent, but to a plain classematic categorization. This can be seen from the fact that the word February does not exhibit a categorial difference in any of the following sentences: (11a) They plan to move in February. (11b) This February is a beautiful month. (11c) The month February I mean is not the month February that you mean 18 .
By contrast, in the following sentences there is a clear categorial difference with regard to the word Cambridge: (12a) They want to move to Cambridge. (12b) This Cambridge is a beautiful city. (12c) The (city) Cambridge I mean is not the (city) Cambridge that you mean.
In sentence (12a), the word Cambridge is a proper name. In (12b) and (12c), on the other hand, the tokens of Cambridge only share the form of the proper name Cambridge but not its function: they are merely homophonous with the proper name Cambridge in (12a). In both (12b) and (12c) each token of Cambridge is an appellative noun, any corresponding monovalent individualization of the referent being prevented by the determiners this (Cambridge) and the [city of ] (Cambridge), respectively. As a proper name, the word Cambridge is inherently determined, which means that there is no need for it to be syntactically determined in a sentence, on the contrary: additional determination inevitably alters its subcategorial status because it automatically implies that the word obtains a classematic meaning. The NP this Cambridge in (12b), for example, is to be interpreted as 'the place/the city/the town ... called "Cambridge"' and the referent is no longer designated in a proprial way but, on the contrary, simply subsumed into a class 19 . Sentences (11a), (11b), and (11c), by contrast, do not contain any such subcategorial alternations, thus revealing the appellative nature of "names" of months.
The difference between appellative nouns and proprial nouns becomes particular obvious if we compare an appellative whose function as a proper name can be independently established in a particular sentence even though its mere form is ambiguous. In the following example, the individualizing function of a proper name is illustrated by the word Friday: (13) Robinson Crusoe was assisted by Friday.
Sentence (13) shows that although Friday, like skat, February, and the other examples in Dummett's list, normally functions as an appellative at the level of langue, it can occasionally be a proper name. In (13) there is no paradigmatic relation between Friday and other proper names from which Friday would obtain its meaning by opposition. Regardless of the basic subcategory the word Friday has as a part of the lexicon, in (13) Friday appears to be a means to individuate a person linguistically without subsuming the referent into a lexical class of the English language. Clearly, a necessary condition for a word to be a proper name is that it is categorially intended as such, in this example as the proper name of a person. However, whether this is the case from a linguistic point of view can only be determined post factum, by analysing actual speech non-taxonomically. In other words, when trying to decide whether a word is a proper name or not, one must not restrict the analysis to a list of existing or possible forms of proper names. Only the function of a "full form" in a given utterance determines whether a word qualifies as a proper name or an appellative noun.

One-dimensionality: The restriction of the proper name to a single referential function
Finally, in his definition of the proper name Coseriu emphasizes that proper names only have one "dimension", as opposed to the two "dimensions" of appellative nouns. This means that a proper name not only individualizes one referent (Raquel) or a group of referents as a single referent (the Bahamas, the Grand Canyon), but that this kind of reference is of a disjunctive kind, the individuation applying to one referent or one group of referents respectively, not both at the same time. By contrast, reference by means of an appellative noun simultaneously covers two "dimensions". Thus, if one refers to All actors on the stage, one means a group of actors as well as each individual actor at the same time. But if one refers to Raquel having a headache the case is different, because now one does not refer to a single member of a group of people all sharing the name Raquel, but to a single individual called "Raquel". Therefore, in a sentence like All the Raquels you know, the word Raquels is not, as Algeo (1973, p. 48-49) erroneously maintains, a proper name. Rather, in the NP all the Raquels the word Raquels is a two-dimensional appellative, the entire noun phrase meaning 'all the persons called "Raquel"' thanks to a semantic reinterpretation of the "mere form" "Raquel" which results in the classematic meaning of Raquels in All the Raquels you know. Likewise, should someone casually or wittily refer to a Bahama, this word too would then function as an appellative, i.e. 'one of the (more than 700) islands of the Bahamagroup' (e.g. Andros). The distinction between one-and two-dimensionality also explains why Dummett's (1981, p. 68-69) contention must be rejected that in the question: the word Cambridge is a proper name. For this question to be understood adequately, the "full form" Cambridge must be reduced to a "mere"-and potentially equivocal-form and subsequently reinterpreted as an appellative ('Which city called "Cambridge"?') 20 . Similarly, if someone refers to his own Grand Canyon on his ranch, then Grand Canyon is an appellative, not a proper name. In all these cases, the words Raquel, Bahama, Cambridge and the Grand Canyon have gone through the stage of "mere forms" in order to function as appellatives on the basis of semantic reinterpretation. Thus the one-dimensionality of proper names as words ("full forms") appears to be intimately linked to the highly idiosyncratic property of proper names that they do not designate the referent as a member of a class. Appellative nouns, on the other hand, are always two-dimensional, and this appears to be a corollary of their basic classematic meaning. The appellative noun denominates a referent or a group of referents and simultaneously represents the denominee as a member of the class for which the noun stands: He is an artist; All artists I know are comedians; This artist was born in Germany. To put it in terms of "dimensionality": with a two-dimensional appellative noun one either refers explicitly to an individual referent and implicitly to the class to which the referent belongs; or one refers explicitly to a group of individuals and implicitly to the individual members of the group. By contrast, a proper name is restricted to a single referential ("individualizing") function 21 .

Meaning, language use, and formal "ipsoflexivity"
The distinctions expounded in this paper between appellatives and proper names prove that proper names are indeed of a genuinely semantic nature, and a highly complex one at that. The difference between appellatives and proper names is not that the former have meaning whereas the latter lack it. Rather, both nominal subcategories have meaning, but their meanings are of different types corresponding to their different functions as parts of speech in actual utterances. Obviously, in both cases this semantic difference does not relate to "mere forms" but to "full forms", and together appellative nouns and proper names constitute the category of "noun".
As a consequence of this analysis, it is impossible to maintain that a word is "used" at times as an appellative, at other times as a proper name. Such a basically taxonomic view inevitably reduces the "full forms" of appellatives and proper names to "mere forms" that are potentially equivocal. The traditional distinction between a proper name and the "use" of a proper name (cf. Pendlebury, 1990) is highly misleading. Given that the subcategory "proper name" (like the subcategory "appellative noun") is necessarily derived from language use, to "be" a proper name means to function as a proper name in actual speech. When scholars frequently refer to "a proper name that is not used as a proper name", they merely abstract from a proper name to its "mere form", reinterpreting it either as an appellative, as in He loves his Cadillac, What a Bush does, a Clinton can do better, etc., or as a citation form only, as in sentences like They called their son "Andrew". At any rate, it is phenomenologically quite inadequate to speak of a proper name that is "used" or "not used" as a proper name 22 .
From the same phenomenological point of view, though, it is significant that proper names often have been interpreted-partly on the basis of intuition, or so it seems-as mere formal signs with specific pragmatic and referential functions, but without meaning proper. One of the most famous theories indebted to this view is the "x called y" theory, which has enjoyed a fair amount of support during the past few decades. Minor differences aside which are due to different scholars' individual approaches, the "x called y" theory amounts to the hypothesis that a name like Aristotle must be analysed as the man called "Aristotle" 23 . According to this theory, the meaning of a proper name is to be found in the formal specification of a referent by means of an implicit proposition is called. It will be clear by now that such an explanation is unacceptable, for a number of reasons. The principal objection is that the "x called y" theory presupposes a classematic interpretation of the meaning of proper names while simultaneously reducing the linguistic 21 Given the complexity of the features monovalence, individualization, and one-dimensionality which I have presented in this section, it is not surprising that there have been several attempts in the history of the theory of proper names to construe the difference between proper name and appellative noun as a difference in degree rather than in kind (cf. Coseriu, 1955a andWillems, 1996, Chapter IV for discussion). This is currently also the approach favoured by scholars who apply prototype theory in their account of proper names. 22 For a more elaborate account of this issue see Willems (1996, Chapters I and II). 23 Cf. Jakobson (1957Jakobson ( /1971, Algeo (1973, p. 212), Kleiber (1981, p. 385), Seiler (1983, p. 151), Kubczak (1985, p. 288), Katz (1994, p. 5), among others. subcategory of the proper name to a label that is itself meaningless. The object to which the implicit proposition is called applies, is indeed a classematic content: Aristotle 'the human being/person/man etc. called "Aristotle"' , Cleopatra 'the woman etc. called "Cleopatra"' , Africa 'the continent called "Africa"' , and so forth. As a matter of fact, Castañeda (1985, p. 107) implicitly deconstructs the "x called y" theory by pointing out that it relates the meaning of the proper name to the fact that "the classifying is done through the calling" (my italics, K.W.). The truth is that a proper name does not classify, even though its function as a part of speech is founded upon "classifying" appellatives. Moreover, by subsuming the semantic function of the proper name under the classifying function of an appellative, the "x called y" theory is based on a false premise, conflating two things that need to be distinguished: i) the rationally secondary status of proper names with regard to the primary appellative nouns, and ii) the definition of the proper name as a genuine part of speech (see §5.2).
Ill-founded as the "x called y" theory may be, it is nonetheless valuable thanks to the attention it pays to the form of the proper name. It is indeed legitimate to relate the semantic function of proper names to the particularity that by means of a proper name a referent is called "y" rather than, say, "z". Unfortunately, this particularity is misunderstood in the "x called y" theory. The undeniable formal peculiarity of proper names was stressed early on by the philosopher Anton Marty (1908) and his disciple Otto Funke (1925) (Coseriu, 1955a, p. 7-8). But it was Jakobson (1957Jakobson ( /1971 who described it in a misleadingly simple way, arguing that proper names have a structure in which the code refers to the code (C/C) 24 . Against the background of the distinctions outlined above, proper names undeniably exhibit a specific "ipsoflexivity" of linguistic form, yet this very phenomenon is in need of further specification 25 .
The formal ipsoflexivity of proper names is intimately connected with the fact that proper names are "full forms", rather than "mere forms". In the speech act (which as far as proper names are concerned is always an act of naming, too, although not of "baptism"!) the form of the proper name is indeed "marked" in comparison to the form of the appellative. The relation of the proper name to the referent is realized precisely in order that the referent is called "y" rather than, say, "z". The appellative, on the other hand, is "unmarked" with respect to formal ipsoflexivity. The goal of the appellative designation of the referent (and of the speaker's corresponding intentional focus, i.e. the specific subcategorial mode of speech) is to categorize the referent "through" the form and the lexical meaning of the word in order to "grasp" the referent as a content of speech (Husserl, 1939(Husserl, /1985, and this does not involve a reference to the code itself in the way it does with proper names 26 . Thus, as far as natural language is concerned, the form of an appellative and its meaning contribute equally in forming a vehicle for designation. By contrast, the form of a proper name is more than just one side of the bilateral sign vehicle: it is itself a goal of the linguistic, viz. proprial, semiosis. The preceding analysis has shown, however, that this property of a proper name clearly is not to be interpreted as a reduction to a "mere form", on the contrary: the formal ipsoflexivity of proper names is nothing but the exploitation of the specific formal nature that is characteristic of linguistic signs in general and has been labelled "full form" in the present paper. What distinguishes proper names from appellatives is the fact that the phonological and morphological side of language is itself functionalized in proper names (and within the nominal subcategory) in a way that is normally absent in appellatives and all other partes orationis.
Finally, it is important to realize that both the formal ipsoflexivity of a proper name and the inherent determination of a proper name (which explains why it does not require determinators such as articles or pronouns to be "actualized" as a referring term in discourse) are proof of the "markedness" of a proper name as a nominal subcategory, not of its seeming "unmarkedness". First of all, in ordinary speech, the form of linguistic signs is not functionalized because it is not itself made an "object" of speech and intentionality in the Husserlian sense (Husserl, 1939(Husserl, /1985cf. Cobb-Stevens, 1990, Chap. VI). Secondly, "primary" appellative nouns do not by themselves refer unless determined syntactically, whereas "secondary" proper names do not need to be "actualized" syntactically (cf. Coseriu, 1955b). Hence, proper names are not characterized by some "lack" or other in comparison with appellatives, but on the contrary by a surplus.

Concluding remarks
There is substantial evidence for the view that most theories of proper names advanced since World War II (for details, see Willems, 1996) have failed to do justice to the relation between form, meaning, and reference with regard to this highly complex part of speech. Neither formalism nor referentialism nor a combination of both suffices to give a complete account of the specific and complex relationship in which form, meaning, and reference are involved in proper names, and mutually condition each other. In this paper, I have presented observations to the effect that the theoretical discussion of the nature of the proper name must be completed by a rigid focus on linguistic semantic. The outline presented has been based on basic assumptions of linguistic phenomenology, especially an analysis of the relation between theory and meta-theory, between intention and intuition, as well as of the nature of linguistic meaning as opposed and related to linguistic form and reference, respectively. The central claim of this paper is that proper names are genuine linguistic signs, and that any theory of the proper name must hence start from an analysis of meaning and referring in natural language. This has allowed me to clarify a number of major issues that need to be addressed by anyone planning to investigate the genuinely linguistic characteristics of proper names in natural language. In this sense, the present paper has been intended as a useful addition to an already fascinating discussion, rather than at challenging existing theories.