Background:

Question 1: Towards the end of the presentation, there is a mention of differences between linguistic competence and linguistic performance. What I have understood is that syntax deals only with linguistic competence, that is, a speaker's implicit I-language that enables her/him to generate grammatical sentences in the given language. Everything that happens after I-Language will fall outside the scope of syntax, and hence they should be handled by other disciplines such as phonetics(sounds production), phonology(speech patterns), sociolinguistics(language use), and so on. Syntax will therefore not go out of its way to talk about how children acquire their first languages and on, but instead, it tries to shed some light on the abstract structure that governs our ability to produce grammatical sentences. From this perspective, syntax is only concerned with linguistic competence and never with linguistic performance. Is my understanding on this correct? Answer: By and large, this is a correct paraphrase. But note that phonological and morphological theory, alongside syntactic theory, are also part of I-language. The reason why the presentation here concentrates on syntactic theory is because this is what the course is about. Also, the theory of language acquisition, at least in part, influences syntactic theory insofar as the requirement of explanatory adequacy demands that the theory of syntax should be compatible with the facts (and their interpretations) observable in language acquisition. Question 2: I have a question about the three types of adequacy (Chomsky 1964). [...] how do they differ from each other? Trying to figure out the difference between the three, a fellow student and I tried to categorize them as follows: 1. Observational adequacy: The theory covers all grammatical sentences of a language (But how without covering structures? Is it just a data set?) 2. Descriptional adequacy: The theory contains the structural representations of all grammatical sentences of a language. 3. Explanatory adequacy: The theory contains the structural representations of all grammatical sentences of a language taking into consideration language acquisition. (But in what way? Language acquisition for children or adult learners, or is this about psycholinguistic processes during speech production?) Answer: Observational adequacy requires that syntactic theory generate the set of linear strings of words that form grammatical sentences (and only these). It makes no statement about the hierarchical structure that underlies these strings. Descriptional adequacy demands observational adequacy plus assignment of the correct hierarchical representations to the strings. Explanatory adequacy demands descriptional adequacy plus compatibility with the observations from first language acquisition by children, i.e., it should offer a solution to Plato's problem: how can the child acquire the complex patters of syntax (and I-language in general) given that the input is very restricted (poverty of the stimulus)? Question 3: I have a question about the poverty of the stimulus in terms of language acquisition. The basic idea of language is that there is a finite set of rules to built sentences. A child starts speaking at the age of around 12 months. So there is a year to just listen to language and the rules it contains. After 18 months it starts to put two words together. As far as I know even the combination of these two words seems to be random (but maybe there is already structure). After 2 years(!) the child starts speaking in small sentences. So after all the baby has 2 years of people around him speaking the language it needs to acquire. Two years to learn a finite set of rules does not seem to be little time. It rather seems that at a certain age, we lose this ability to think abstractly and therefore people like Genie etc. cannot aquire language anymore. Did I understand something wrong? Answer: Maybe two years does not seem little time. I think the intuition is that it really is not much given the task that the child has to master. For instance, it needs to be able to identify word boundaries, a problem we have not talked about at all. It has to filter out ungrammatical utterances that show up in the input. It acquires generalizations that are underdetermined by the input (poverty of the stimulus). And it does so all by itself, i.e. without being instructed. If what you are saying is that -- given the childs capacity to "think abstractly" -- two years are more than enough, then I think that one may agree (or one may not). The purpose of the .discussion was to highlight a) that the child must have some capacity of this kind (which, presumably, is not part of its general "intelligence") in the first place, and b) that this capacity is sensitive to syntactic structure that goes beyond linear order.

Constituency, Projection, C-Selection:

Question 1: On page 16, there's the following mention: "Aside: In (21-b) John is linearized to the right of the predicate killed, which is perhaps surprising (cf. John was killed ). The justification for this will come later." I [...] did not understand what was surprising or what was justified. Answer: I thought that it might, perhaps, be surprising here that John in *By Mary was killed John is placed to the right of the predicate killed, and not to its left, as in the grammatical example By Mary John was killed. Given this, one might think that this example is ungrammatical not because the subject position is filled by the (prepositional) constituent by Mary (instead of by some noun-like constituent), which was the main issue of the discussion at this point, but because John is somehow put in the wrong position. However, as will become clear later (and this is what I meant by justification) when we encounter the Uniformity of Theta Assigment Hypothesis (UTAH) John is in fact merged in exactly the right position in the example under discussion. Therefore, its ungrammaticality needs to be acounted for otherwise (namely by reference to the assumption that, at least in English, the subject position must be filled by a noun). Question 2: You mentioned It rained as an example of a zero-place predicate. Are there any other instances of zero-place predicates in English? What would you say are the main differences between a zero-place predicate and a one-place predicate? (E.g. It rained versus John slept.) Isn't rain an intransitive verb just like sleep is? Answer: In general, weather-verbs are assumed to be zero-place predicates: the it in a sentence such as It rained is not a semantic argument of the verb. To see this you can ask yourself what it could possibly refer to in this context, and what kind of theta-role it might bear. Then you will realize that it does not have any reference, and that it does not bear any theta-role, unless you assume that it = some god-like being (which, crucially, is not the interpretation the sentence has). In contrast, John in a sentence such as John worked certainly is an argument of work: John refers to the person named "John", and it receives the theta-role agent of the predicate. Question 3: On page 23, there is the following example: (27) a. *Dr. Brumm shaved. I feel that shave can also be an intransitive verb as in I shaved yesterday, which makes perfect sense. Was the above-mentioned example deemed ungrammatical because the transitive/two-place predicate aspect of the verb was focused on? In relation to this question, if a verb can either be intranstive (one-place predicate) or transitive (two-place predicate), how would syntax tell us which features of a particular verb we should be looking at? Answer: You are absolutely right. In English, shave does not need its semantic argument that bears the theta-role theme be realized by a syntactic argument when it is interpreted reflexively. In other words, Dr. Brumm shaved is grammatical under the interpretation "Dr. Brumm shaved himself.". And perhaps it is even grammatical under the interpretation "Dr. Brumm shaved someone" (where it is not further specified who he shaved). It seems that many verbs (in English, at least) allow for not realizing the theme-role of a transitive verb, resulting in some unspecific reading (and I have nothing to say here as to why this is the case). I changed the example, which now involves the predicate demonize (following Adger 2003). I hope that this makes things somewhat clearer. And you are also right in posing the question how the fact that a semantic argument needs not be realized is to be analyzed in terms of C-selection. The easiest (but perhaps not very insightful) way to deal with this is to assume that some C-selection features are optional. Thus, a predicate such as shave may come in two versions: one that bears two [uN]s (each associated with one of its theta-roles), and one that only bears one [uN], which is associated with its agent-role.