 Hello. I welcome you to an e-lecture of a series of e-lectures about constituent analysis. As you know, native speakers intuitively recognize that the words within sentences are structured into successively larger syntactic groups so-called constituents. Let us illustrate that using a standard sentence. In this boy will speak very slowly to that girl, a sentence from Andrew Radford's 1977 book Transformational Syntax. We will probably all agree that very goes with slowly and not with speak. We will furthermore agree that this and boy go together and similarly that and girl. And we will also share the opinion that too and that girl somehow build another constituent. And eventually we will intuitively understand that speak very slowly, speak and very slowly belong together. And at a later stage we will group speak very slowly and to that girl into another constituent and eventually we will come up with this sort of hierarchical structure which then displays the entire sentence. So this is the sort of constituent analysis but why do we need to know about such constituents in the first place? Why are they interesting? And secondly, well that's the second question. How do we arrive at such decisions? The answers to these two essential questions in syntax constitute the focus of this e-lecture. Let's start with question one. Why do we bother about syntactic units larger than words? Well, take the sentence John saw old men and women. This construction can be interpreted in two ways. Either John saw old men and women of any age or John saw both men and women who are old. This difference can well be represented by means of a hierarchical representation where specific constituents can be worked out. In the first case where just the men are old we will probably build an independent constituent that links old with men. And this can be conjoined with and and women at a later stage. In the second case where everyone is old well we have men and women that build somehow a constituent and at a later stage they are conjoined with old. So cases like these which are referred to as cases of structural ambiguity can most adequately be represented by means of a constituent analysis. But how do we arrive at such constituents? Well, there are several interacting arguments so-called tests that lend support to the postulation of constituents. These arguments can be subdivided into non-syntactic arguments or tests which come primarily from morphology and semantics. Distributional tests or distributional arguments that is arguments that consider the distribution of constituents within sentences and other arguments that apply specific syntactic operations to the constituents in question. Let's take a closer look and start with the non-syntactic arguments. In many languages morphological considerations provide support in favor of constituents. The assumption is simple. Particular morphological features such as number, gender or case are not exclusively properties of words but of phrases. So we have this sort of relationship. In present day English an almost analytic language this is often only visible in terms of number whereas in German number, gender and case have to be shared within phrases or take languages such as Chinese. This is an interesting case. In Chinese the noun determines the choice of its classifier depending on its semantic potential. So for example book is combined with the classifier ben whereas ball is combined with the classifier g and both are equivalent to present day English piece of. So the argument here is simple. It's not just words but larger constituents that are associated with particular morphological or semantic features yet the most important arguments in favor of constituents are syntactic. In fact many of these syntactic arguments relate to the distribution of various sequences of words that is to the question whether a particular sequence of words has the same distribution as another phrase of the same type. Three types of distributional tests can be distinguished. The preposing test that is we move items to the front and test whether the result is grammatical. The post-posing test we move items to the back and test whether the results are grammatical. And the sentence fragment test where we isolate parts of sentences and test whether they can stand alone. In summary distributional evidence suggests that whole constituents are not just parts of constituents are in the focus of these processes. So let's look at preposing first. Preposing means that under certain stylistic conditions for example to achieve a particular stylistic effect parts of a sentence may be moved to the front. However only a whole constituent can be preposed in this way not just part of it. Here is an example. I don't like your new neighbor. Let's now pre-pose several items. For example your new neighbor and this clearly works. Your new neighbor I don't like. If we pre-pose new neighbor well then the result clearly is ungrammatical. Your new doesn't work and neither does just your. So in other words since only your new neighbor works we have reasons to believe that it is a constituent so it is only the entire constituent your new neighbor that can be preposed and not just parts of it. Similarly we can move elements to the back again to achieve specific stylistic effects but again only a whole constituent can be post-posed in this way. Let's demonstrate that. He explained his problems to her and let's find out what we can post post. For example we can post post his problems. He explained to her his problems but we cannot post post just problems. Neither can we just post post his. So these latter two cases are ungrammatical and again it is only the entire construction his problems that can be post-posed not just part of it. Another distributional test has become known as the sentence fragment test. In a large number of contexts especially in conversations short form replies are made. These replies which are also referred to as sentence fragments are highly constrained not surprisingly only whole constituents can serve as sentence fragments. So imagine the following question why don't you who don't you like well the answer could easily be your new neighbor but you wouldn't use new neighbor as an answer that's ungrammatical or take this one what did he explain to her well of course his problems that's legitimate but if you just use his as an answer well that's ungrammatical so again it's the entire constituent that has to be used as a sentence fragment. Still there are other types of syntactic evidence that support the claim that sentences are structured out of constituents. For example the coordination test which tests constituent identity or the proform test that tests on the basis of a replacement tests that replaces constituents by appropriate proforms or the ellipsis test which tests the possibility of dropping or omitting constituents. Let us look at coordination first present day English like many other languages has a variety of coordinating conjunctions and but or well we could also use things like however and so on they are used to coordinate to conjoin words or phrases interestingly we cannot conjoin any randomly selected elements but only categories of the same type so x and y is impossible but x and x would be possible let's exemplify this John wrote to Mary is one of our examples which i'm going to use to exemplify this and John wrote a postcard now obviously we have two constituents we can easily show that they are constituents to Mary John wrote a postcard John wrote both can be proposed for example but if we conjoin them John wrote to Mary and a postcard the sentence becomes ungrammatical so here we have a nice diagnostic test for the status that is the position in the hierarchy of constituents obviously to Mary and a postcard are constituents but they're different ones let's illustrate that further here are two constructions starting within John punched Bill in the face and John punched Bill in the park the fact that John punched Bill in the face and in the park is ungrammatical tells us that in the park and in the face must have a different status that is they're different within the hierarchy of a sentence they are constituents but they're different another test for constituents is the proform test proforms are elements that refer to other items the most commonly known elements in this respect are pronouns as we will see in a second they do not just stand for nouns but for larger structures so-called noun phrases here is an example John punched his friend in the park let's now apply the proform test here are the pronouns John can be replaced by he and his friend can be replaced by him so they're both constituents or take in the park in the park can be replaced by there so in the park is a constituent in fact we call it a pro prepositional phrase form there is a pro prepositional phrase form and in the park is a prepositional phrase and what happens if we ask a question such as did John punch his friend in the park well the answer could be something like yes he did in which case did would stand for the entire predicate referred to as pro verb phrase so proforms are important substitutes for constituents let us finally look at ellipsis and a certain discourse conditions it is possible in English and in many other languages for some part of the sentence to undergo ellipsis that is to be omitted provided that the omitted part can be recovered from the context here is an example John won't solve his problems well and like in all the other tests it is the entire constituent solve his problems that has to be omitted not just I bet he will solve just the his problems the last part so ellipsis is something that predominantly applies to predicates now that we understand how we can find out whether one or several words can be grouped into constituents and that we have understood that sentences have a hierarchical structure we can apply these criteria however often not all of these criteria apply for example if a string of words is already at the end of a sentence he sat in the park how can we postpose it or take the item V in the park here morphological criteria do not apply since the is neutral with regard to number the park or the parks no difference nevertheless there are always some criteria that can nicely help to define the status of constituents try these two sentences the dog barked in the park and he took up the phone are the strings starting with in in the first case of the colored strings and up the phone are they constituents or not just take your time and pause the video the solution will come up in a few seconds okay and here is the solution since the coordination test the first test in the park and up the phone results in an ungrammatical construction the two strings in the park and up the phone must be different they must have a different status and since the remaining tests ellipsis proform and preposing only work for in the park as in the dog barked the dog barked there or in the park the dog barked since this is the case in the park is a constituent but up the phone in the sentence he took up the phone is not let's summarize now that we understand the mechanism of constituent identification we have to perform the next step and find out the types of constituents that are part of today's theoretical grammar perhaps you already know that there are noun phrases verb phrases propositional phrases and even tense phrases so whenever we postulate such constituents and even intermediate constituents such as n bar or verb bar you heard about them well whenever we do that we will come back to one or several of the tests we discussed in this e-lecture so make sure you not only remember these tests but more importantly that you always know how to apply them so see you again in one of my e-lectures about constituent analysis thanks a lot for listening and bye bye