 The definition of the principles that define the legitimacy of lexical insertion constitutes the program of this e-lecture. First, we will specify the problem using several examples. After categorizing the lexines that are involved, we will first specify their local syntactic contexts. This is referred to as sub-categorization. Finally, we will add thematic aspects that expand the mere syntactic definition of lexines. Let us start with a syntactic tree that is generated by the phrase structure rules and now waits for lexines to be inserted into its terminal nodes. Here is the tree and this is our lexicon. It contains some noun phrases and some verbs to make it as simple as possible. Now, let's first of all generate a sentence. Let's say we generate a sentence where we move C into this node, where we have the table over here and where the noun phrase John is inserted into this position. Now, such a sentence is not possible at all. It violates the principles of present-day English word order and it can be ruled out relatively easily because it involves a categorical mismatch here. C is not a noun phrase and the table is not a verb. So, by a simple principle that compares the terminal nodes with the categories that are associated with the lexines, this sentence can be ruled out. Let's look at another example. Here we have the same syntactic tree but let's now insert the following. John, now here we have categorical match. John is a noun phrase. Give is a verb so this would be John gave. Well, and then something like John gave the book. Now, we clearly have categorical identity. The categories do match but there's something wrong with the syntactic context. Give requires two objects but there's only one so this is the object of give. Thus, the ungrammaticality is due to a violation of the local syntactic context of give. Let's take a further example. The table is clearly a noun phrase. C is a verb past tense. Here is the past tense node. Well, and let's say the table saw John. Now, everything is right now. However, there's something wrong with the meaning of the verbal arguments. The table cannot usually be the subject of C. Thus, the argument structure of C is violated. Let's summarize these observations. Lexical insertion is determined by the following principles. Now, in the first case, we have been observing a categorical mismatch. In the second case, the local context was violated. So, here we have a problem of local syntactic context violation. And in the third case, the table saw John. We have there was something wrong with the argument structure of the verb C. Well, and there are three principles that guide or that help us to solve these issues. In the first case, we have the phenomenon of categorization. The definition of the local syntactic context is referred to as subcategorization. And in the third case, we have something that is referred to as argument structure. Let's start with categorization. Now, this is very simple. In the lexicon, all lexemes are listed with their syntactic category, that is, with their word class. So, a book is then, and now normally you present the lexemes in capital letters, then a colon follows, and then the word class is indicated. V is a determiner, give is a verb. About is a preposition. PST is a short term for past tense. And in generative grammar, it is, it constitutes part of the inflectional node. Well, and round in the sense of a round ball is an adjective. But please note that, of course, elements like round can have several word classes. So, for example, round can be also defined, for example, as a noun in the sense of a round of beer, or it can be a preposition as in round the house. So, all lexemes are represented with their syntactic category in the lexicon. Let's now look at subcategorization. Now, subcategorization defines a lexical category within its local syntactic context. Here is an example. Let's take the verb C. Now, by convention, subcategorization facts are represented like this. First of all, a comma separates the categorical information from the subcategorization facts. The subcategorization facts are represented in square brackets. As follows, first of all, you indicate the position of an item. That is, that is in this case C by an underscore. And then you specify the local syntactic context by means of the obligatory elements. That is the complements that follow or precede. In this case, we know that C requires a noun phrase in its local context as in John saw Mary. Now, this notation is referred to as subcategorization frame. Let me abbreviate that to subcat frame. And it helps us to generate the local context and to specify it as a tree structure. All you have to do is you take the category of the item that you specify and the local context and define a new mother node, which in this case is verb bar. So this is a local tree that can be directly generated from the subcategorization frame and the category of C. Let's look at another example. Here we have an adjective, fund. Now, a possible context could be fund of someone or something. Now, the subcategorization frame looks like this. Here you have the position of fund. Now, the element that follows is, in general terms, a prepositional phrase, fund of someone or something. Well, and here is the local tree again. Quite simple, isn't it? We could say a little bit more about the prepositional phrase. We could, for example, say that the prepositional phrase consists of a preposition of and a noun phrase. So this is a further subcategorization, a piece of subcategorization information, which then already indicates to us how the prepositional phrase is to be constructed internally. Now, the last component we have to observe or the last aspect we have to observe is the argument structure of an item in the lexicon. Now, we know that the specification of lexical entries via categorization and subcategorization alone is not sufficient to prevent sentences such as the table saw the woman from being generated. In order to overcome these anomalies, it was suggested that the arguments of a lexeme have to be included in the syntactic specification of the context of an item. In the early days of generative grammar, that is, in the standard theory of 1965, this was achieved by the definition of so-called selection restrictions. Well, let's write down the word here just to remind you of this word. Well, you might want to Google it, selection restrictions. That is, by means of semantic features that were associated with the arguments of an item. For example, C would then be specified as a verb that requires a subject that is animate, plus, minus, animate. This approach was generalized in the 1980s by so-called thematic grids. Now, this here is a thematic grid. That is by a more general specification of the arguments of an item. This proposal goes back to Charles Fillmore and his famous publication, A Case for Case, in the 1960s, where he defined the thematic structure of, for example, verbs by means of so-called deep case relations. They were later redefined as thematic relations and were integrated into the theory of generative grammar. The most well-known thematic relations are, for example, things like agent, that is, the instigator of an action. Patient, that is, an argument that describes the suffering sort of action. Or, for example, a theme. Well, this would be an argument about which a certain action is. Now, today these thematic roles, or in short theta, roles, theta, the Greek letter theta, roles are associated as follows with the items in the lexicon. We know, for example, that C has two arguments. A subject, which is normally a noun phrase, and an object, which is normally a noun phrase. Give, by contrast, has three arguments. Someone gives another person something. So, you have three noun phrases, if you wish. You could also have a second version where you have a prepositional phrase. Instead, someone gives something to someone. Now, what about the theta roles? Well, in C, the first subject or the first argument must be an agent. Someone sees. What do we see? Well, this theta role is normally a theme. So, we see something. Something about which the seeing is. Give is similar. The subject argument must be an agent. Someone who gives something must be an instigator of an action. Well, we give someone something. So, this is then perhaps the goal. John gave Mary the goal, a book, and a book. The last argument would be a theme again. Now, let's summarize. The combination of categories, subcategories, and thematic information, specifies the syntactic, and to some extent, the semantic default context. So, this is important that these things specify defaults. The default context of leg seems. However, as usual with language, there are always cases where the defaults have to be overwritten. So, there's a big but. Let's take a sentence like, well, I'm not writing it down here. Yesterday saw a smiling president. There, of course, the subject argument of C is by no means an agent. Or if you take another example with give, the candles gave the city a wonderful atmosphere, then the candles clearly are not agents. So, do we then violate the thematic grid of C and give? Well, whether this is an accidental violation or a systematic principle of present-day English, can be found out if you consult our e-lecture subjects in present-day English.