 What is the relationship between a word and an object and how can we define this relationship? The branch of semantics concerned with this question is referred to as referential semantics or simply reference and it constitutes the focus of this e-lecture. In particular, we will look at types of reference. We will deal with the naming view and more modern approaches towards concepts in the human mind and we will take a look at prototypes. But let us first of all locate reference within the overall field of semantics. Within semantics, with its main branches, word semantics, sentence semantics and utterance semantics, word semantics seeks to explain the phenomenon of meaning in natural language by means of defining the nature of word meaning. There are two ways of looking at word meaning resulting in two branches. Reference or referential semantics and sense or lexical semantics. We will concentrate on reference in this e-lecture, that is the definition of the meaning relation between words and objects. That there is clearly such a definable relationship between words and objects is undisputed. We use words to refer to objects. For example to the object fly, which is defined as an animate physical object, an insect with six legs and two wings. Or we use words to refer to actions such as cooking. So reference is concerned with designating things, activities, properties, relationships and so on in the outside world by means of linguistic items. The relationship between a word and a class of entities it stands for is often called denotation. A number of semantics make a distinction in their terminology between two terms, where denotation is used for the relationship between linguistic expressions and classes of entities in the world and reference labels the activity of picking out precise things of reference in the world on some specific occasion. In short, referring is what speakers do while denoting is a property of linguistic expressions. The following types of reference can be defined. Definite reference, indefinite reference and generic reference. Let us look at these types of reference in detail. Definite reference is an act of identifying entities through linguistic expressions, which is characterized by the following properties. Here is an example. The doctor sent it to her. Now the noun phrase the doctor can be uniquely identified by the speaker and the listener even though they may not have full information about the referential target. But they both implicitly assume that it is enough to pick out the intended referent. The set of referring expressions in English includes these categories here. For example, definite determiners such as the or a determiner such as our. Pronouns such as I or she names such as Paul or Mexico. And then we have locative adverbs such as here and there and temporal adverbs such as now or yesterday. Let's look at indefinite reference next. In instances of indefinite reference, the success of the act of reference does not depend on the individual features of the referent. Only the features of the class to which the referent belongs are relevant. Here is our example again, slightly modified. Now the noun phrase a doctor is not necessarily identifiable neither by the speaker nor the hearer. In fact, the identity of the reference is not relevant to the message. To understand the speaker, the hearer does not have to know which doctor is involved. It is not relevant whether either speaker or hearer are actually able to uniquely identify the reference. So it's just any sort of doctor. Indefiniteness is signaled by the following elements in present day English. For example, by indefinite determinants such as a or many indefinite pronouns such as, well, let's take one of them, anyone, anywhere would be a locative adverb and a temporal adverb would be something like some time. Okay, let's look at generic reference next. Now there are instances where potentially referring elements such as nouns do not pick out a particular entity but refer to a class of reference. This is called generic reference. Here is an example. Now, all three sentences, the computer is a valuable tool, the computer is a valuable tool and computers are valuable tools. All three sentences have readings which involve generic reference. Each of them states that being of great use is a general characteristic of the members of the class of computers. Sentences involving generic reference may make statements about the whole class of reference. And then it is called collective reading. For example, the computer is obsolete. And they may make statements about each member of a particular class. For example, a computer is obsolete, a computer has a monitor and so on and so forth. This is called distributive reading. Having discussed the types of reference, the central question arises, how can we define these various types of reference? One of the oldest views can be found in Plato's dialogue cratilos, the so called naming view. It takes the position that the word is a name for an object that is meaning is reference. This works with proper descriptions where we clearly have a word that identifies a particular object in the real world. However, a number of difficulties with the naming view indicate that meaning has to be different from reference. These difficulties concern abstract objects such as love, extinct objects such as dinosaur where we do not have a reference anymore, objects in the real world that have several names such as the planet Venus which can be referred to as morning star or evening star. And then opaque context such as John knows that Bill wants to kiss Mary. Thus we need a more sophisticated view that relates words and entities. Modern semantics has adopted a more complex perspective in which reference is only one dimension of meaning. The other dimension is referred to as sense. The German logician Gottlob Frege, who lived from 1848 to 1925, was the first to distinguish two dimensions of our semantic knowledge of an expression, its sense and its reference. Frege used the German words Zinn and Bedeutung respectively. The modern solution to the problem of explaining the nature of word meaning is to define the meaning of a word in terms of the image or concept in the speaker's or hearer's mind. This relationship can best be illustrated by means of the Semiotic Triangle as introduced by Charles Ogden and Ivor Richards in 1923. The idea is that words and objects may be related in several ways. One way is a more or less direct relationship. It can be realized in two ways. One possibility is Onomatopoeia, that is a relationship between a linguistic sign and an object via its phonetic shape. Thus a direct relationship between sounds and objects. Onomatopoeia is Greek and consists of the following components. Onoma, which means name, and pier means what. It works well with items such as cow, where moo is used almost all over the world. However, it works less well for dogs, where English dogs go bow-wow, and German dogs would go vow-vow. Well entities is even less comprehensive if we look at the cock over there. In German it goes kikiriki, in English cuck-a-doodle-doo, and what about your language? I think you can clearly see that onomatopoeia only affects a handful of items. This is why this bottom line here is dashed and not a straight line. This indirect relationship can easily be confirmed by a second relation that is often said to be more direct. It is referred to as iconicity, that is the use of imagery in linguistic signs. Or more precisely a term referring to the property of some linguistic symbols whose physical form closely corresponds to the characteristics of the object they refer to. Here are two examples where it seems to work. Now, big items such as this object here are often produced using words that involve a large vocal tract, a large aural opening, grand, gross, boar, which means big object. And tiny objects like this mouse often involve very much compressed vocal tract configurations where you have a small opening, tiny, petite, klitzeklein. But as you all know it doesn't work all the time. Take the German word riesig, which involves a small vocal tract, but means big. Or take the word big itself, it doesn't involve a big vocal tract. So in human language, as opposed to animal communication, only a few items are said to possess such symbolic properties. In the majority of the cases, the relationship is established via concepts in the human mind. The problem here is to know what form these concepts take. Concepts have the status of mental categories because they classify experience and provide knowledge about entities which fall into them. But what is the best way of characterizing conceptual categories? The classical approach defines a conceptual category such as animal or man in terms of a set of necessary and sufficient conditions for an entity in the world to be a member of that category. By the way, by convention concepts are normally written in capital letters. Despite its intuitive appeal, this approach has a number of shortcomings. The meanings of concepts such as love or house cannot be captured by means of necessary and sufficient conditions. There are too many variants and it is too abstract. A description of a category in terms of conditions implies that it has a sharp fixed boundary. But as various experiments have shown, the boundaries of categories such as warm and cold are fuzzy and change with the context. Well, and then the approach cannot explain why some members are felt to be better examples of a category than others. For instance, most of you would agree that penguin is not the best example of a bird. It cannot fly. Due to these problems, several more sophisticated theories of concept representation have been proposed. One of them is the theory of prototypes. Prototype theory assumes that conceptual categories possess an internal structure comprising of central members, less central members and borderline cases. So, for example, a penguin is a borderline case as well as the kiwi. Both cannot fly. Whereas birds such as a robin, a sparrow or a thrush are very prototypical. Well, an owl is perhaps somewhere, well, let's put it over here, in the middle. Category membership is defined by reference to a cluster of attributes that characterize the prototype. Members that are closer to the prototype share more features with it than more peripheral members. A conceptual category is thus structured by a network of similarities which unites its members, centered around the prototype. The boundaries of a category are not clearly defined, but fuzzy. Now, despite its success, the prototype theory is not without problematic aspects. For example, the exact definition of the properties that identify the prototype is still very much disputed. Now, the final issue we have to sort out is how are words mapped onto concepts? Well, linguistic expressions, words map onto complex multidimensional networks of concepts. The links between the concepts are of different kinds, is a kind of, is a part of, is used for and possess variable strengths. Words directly activate concepts and indirectly give access to other concepts linked to them as well, depending on the strength of the link. For example, the word dog has a direct link to the concept dog, but only an indirect one to the concept, to the superordinate concept animal. There are several possible configurations for the mapping between words and concepts. There are one-to-one relationships, aircraft vehicle. There are one-to-many relationships where table activates two concepts, a piece of furniture and a set of data. Or many-to-one where words such as kill, murder, assassinate, dispose of and liquidate, activate the concept cause to die. Many-to-many is also possible as a mapping, but it results from combining the one-to-many to the many-to-one relationships. Let us summarize. This e-lecture cannot handle all aspects of reference. However, we have shown that the relationships between words and extra linguistic objects are far more complex than expected. One question was not addressed in this e-lecture, the question of how we can formalize referential relations. We will discuss this issue in our e-lecture theories of word meaning, but prior to that I will introduce you to the main implications of lexical semantics and the central sense relations. So, stay tuned.