LECTURETTE #18: CASE IN MINIMALIST THEORY I have so far pretty much managed to avoid discussion of Minimalist Theo- ry, the most recent version of the Standard Theory. I will probably live to regret this; the more i think about it, the more it seems that i've passed up some good opportunities to introduce some of the peculiarities of this perspective on syntactic theory, and I will therefore have to do some backtracking in the near future. Be that as it may, i have now reached the point at which the issue(s) become(s) unavoidable, since Mini- malist Theory involves a radical departure from the view characteristic of REST up through the mid-80's on the question of Case-Assignment or, more properly, how NPs manage to pass the Case Filter. The most basic and, at least in some ways, important difference is this: in Minimalist Theory, *nothing assigns Case* *Case Assignment, properly so-called, doesn't happen* -- at least, it doesn't happen in syntax. NPs don't get Case assigned to them by their syntactic governors. They come into the syntactic struc- tures they inhabit already bearing Case. You'll remember that in the last Lecturette i mentioned Marantz' notion that morphological case-markers (in languages that have them) can assign abstract, syntactic Case as well as overt, morphological case, and that to the best of my knowledge this hypothesis has never been seriously challenged within orthodox PPA. Well, in modern Minimalist Theory this notion -- like the whole notion of 'abstract' Case itself -- has been extended to all languages: Case is something that nominals bring with them out of the lexicon into syntax, along with their overt phonological rea- lizations (whether these include symptoms of Case or not), their semantic and syntactic cooccurrence restrictions if any, etc. The action of the syntactic component is limited to establishing whether the particular Case a nominal bears is licensed in that particular structural position or not. This is done by 'feature-checking', which is probably the real core of Minimalist Theory; at any rate, it's the area that seems to be driving the most research nowadays. Well, if NPs already bear Case from the point of their initial appearance in the syntactic structure (i hear you ask), how can the Case Filter motivate movement? Especially given the three fundamental principles of the Minimalist Approach: LAZINESS: A constituent doesn't move unless it absolutely has to (e.g., to satisfy the Case Filter or some other grammatical constraint) PROCRASTINATION: A constituent doesn't move any earlier than absolutely necessary (i.e., movement will occur by SS only if it's required by con- straints at both PF and LF) GREED: A constituent doesn't move to satisfy a constraint that properly applies to another constituent (e.g., just because it would make things more comfortable for the verb if a certain NP moved in a certain way is no reason for that NP to move) Given especially the principle of Greed, it would seem that assuming that NPs come into syntax already bearing Case would gut the Case Filter's power to motivate movement. For instance, assuming that the subject NP is base-generated as Spec of VP, and already from that moment bears No- minative Case, why should it move to Spec of S? The answer in the Minimalist Program has to do with licensing. The base position (DS position) of an NP licenses its theta-role but not (necessa- rily) its Case. (I don't think i can emphasize this enough: We must keep clear the distinction between theta-roles, basically a semantic notion, and Case, basically a syntactic notion with occasional morphological and semantic repercussions.) If its Case is not licensed in its base posi- tion, then it must move to a position in which that Case is licensed. Typically, this is the Spec position of some functional head. Another way this is often described in the literature is that when an NP is base-generated, it doesn't merely bear its syntactic Case (which may or may not be 'visible', in the sense of being marked by overt morphologi- cal or lexical material), but also abstract Case 'features' WHICH MUST THEMSELVES BE ELIMINATED before the derivation of the clause is complete. (Most versions of the Minimalist Approach that i've come across assume that what's critical about the abstract Case features is whether or not they are 'visible' at a particular level. It's not enough for them to exist. What 'visibility' consists of is not entirely clear, and seems to vary from one researcher to another. It isn't necessarily a function of phonological visibility in the old sense of being manifest in the form of overt phonological material, though in some versions there seems to be a connection. It may also involve some non-phonologically-overt garbage that PF, say, is nevertheless sensitive to. Imagine something like an NP arriving at PF without any overt case-marking but surrounded by maledicta -- those typographical doohickeys (e.g., #@%$*!!) traditionally used in comic strips to represent linguistic expostulations that are unprintable for reasons of social decency. Bear in mind: *overt case-marking* is fine and can get past any version of the Case Filter; *abstract Case fea- tures* are not and have to be stripped away. But occasionally one comes across suggestions that the Case features may in a certain language be so abstract that PF, or LF, or whatever doesn't even notice them, in which case they don't need to be stripped away. I don't know if anyone has se- riously proposed such things in the analysis of any particular language, but if so then NP-movement would presumably not be characteristic of that language, at least not to satisfy any version of the Case Filter.) Think of it this way (better analogies may be possible, but this is the best i've been able to come up with so far). Imagine that Case is like a suit of clothes. You need to wear it in order to be presentable in a given social situation (hence the Case Filter). But it comes to you com- plete with the ridiculous plastic bag the dry-cleaners always use, and you can't actually wear the clothes until you've gotten rid of the plas- tic bag. So you've got to dump the bag somewhere. That 'somewhere' is the Spec position of the appropriate functional head. Note the word 'appropriate' in that last sentence. It is assumed that a given functional head is empowered to check and eliminate Case-features for only a particular Case or range of Cases for its NP-Specifier, not any Case at all. Thus, Agr(S) is empowered to license nominative Case, or the language-particular equivalent thereof, not any other; Agr(O) is empowered to license only those Cases that belong to the range of 'objec- tive' Cases (e.g., accusative and dative). This is why (assuming for the moment, contra Thrainsson and van Gelderen, that the functional hierarchy is universal and all languages exhibit the same set of functional heads, in particular that English grammar includes an Agr(O)) Agr(O) is able to license objective case on the pronoun in (ia) but not subjective case as in (ib). (i) a. Gertrude hid me behind the arras. b. *Gertrude hid i behind the arras This view of the matter assumes that for any Case recognized by the gram- mar of a particular language there must be at least one functional head generatable by the syntax thereof that will license, check for, and eli- minate that Case and the abstract features associated therewith. More on the problems with this notion below. OK, so you've got your Case, brought it with you from the lexicon (actual- ly, of course, only the head of the NP gets Case-marked in the lexicon; the rest of the NP takes its Case from its head, which means that any other constituents in the NP that have to agree with the head in case -- adjectives, determiners, etc., depending on language-particular defini- tions -- have to manage somehow, but that's another story). You've also got, along with it, this irritating bundle of abstract Case features -- the plastic bag. At what point do you get rid of that? Remember the Principle of Procrastination: you put off getting rid of it as long as possible. (This is where the analogy breaks down. For myself, i always want to get rid of the plastic cleaner's bag as soon as possible.) And remember what i said about the Principle of Procrastination: Movement occurs by SS only if it's necessary to satisfy constraints at both PF and LF. That's because, as you'll remember, derivational history branches at SS; PF and LF are both derived directly from SS but neither is derived from the other. Therefore, if a certain movement is required by both PF and LF constraints then it must take place by SS even if it isn't re- quired by SS constraints. The older version of the Case Filter was a PF-constraint, and during the 80's it was assumed that NPs were checked for compliance with the Case Filter at PF. More recently, it is assumed that the abstract Case fea- tures, *should they be visible at PF*, need to be eliminated before that stage in the derivation is reached. This requirement would be met by movement to an appropriate Spec position that occurs during the deriva- tion of PF from SS. But it is also conceivable that the abstract Case features might be visi- ble at LF, or (alternatively) that LF also, for reasons of its own, might require Case-marking to be licensed. (Typically, such a requirement is spoken of in terms of making sure that the Case-marking is compatible with the theta-role an NP bears, otherwise the semantic interpretation of the clause breaks down. While as hinted earlier there is no one-to-one correspondence between theta-roles and Case, there is generally some kind of restriction on the range of Cases that can encode a particular theta- role. In other words, it may not be the case that a given theta-role is always borne by an NP with Case A; but it is very typically the case that a given theta-role is always borne by an NP whose Case-marking is limited to a small number -- usually two or at most three -- of the options al- lowed by the grammar of the language in question.) If such an LF require- ment exists, then the NP in question must move to the proper Spec position by LF; if the relevant abstract Case-features would also be 'visible' at PF, then such movement must also occur by PF, in which case it must occur in the derivation of SS. So we have three options, amongst which the languages of the world presumably choose on the basis of various parame- ters: (1) Abstract Case-Features visible at PF LF doesn't care about them ergo, 'stylistic' movement at PF (2) Abstract Case-Features invisible at PF LF sensitive to Case-checking ergo, invisible movement at LF (3) Abstract Case-Features visible at PF LF sensitive to Case-checking ergo, movement at SS. (The fourth option, in which no movement is needed and therefore, by the Laziness Principle, does not occur, is theoretically possible but as i said above as far as i know has never been proposed for any actual language.) So far, i've been talking mostly about movement of major argument NPs into the Spec-positions of various functional heads such as Agr(S) and Agr(O) for the purpose of Case-checking. But according to the older version of the Case Filter all phonologically overt NPs require Case, and there's no question that in languages like Latin, Sanskrit, and Warlpiri that have rich overt morphological case-marking systems every single overt NP has case (and therefore presumably Case) whether it's a core argument or an adverbial or what. If the Minimalist approach to Case is valid, presumably the NPs 'Terry' and 'fountain pen' in (4) also bear abstract Case-features. Where do these NPs go to get their Cases checked and their abstract Case-features dumped? In the English sentence, both of these NPs are, at least superficially, the complements of preposi- tions, and in the older version of Case Theory those prepositions would be held responsible for getting their respective complements past the Case Filter. But the Minimalist approach presumably requires the NPs in question to dump their abstract Case-features in the Spec position of some functional head, and i have yet to hear that adpositions have func- tional heads associated with them. (Vincent deCaen of the University of Toronto has recently posted a query on LINGUIST about movement to a Spec- of-PP position. I'll be interested in seeing his summary of responses, should he ever post one.) (4) Leslie wrote the letter to Terry with a fountain pen. Furthermore, there are other languages in which the equivalent NPs would not be complements of adpositions, merely marked with one or another of a variety of oblique morphological cases. (For instance, i can say in com- plete confidence that any attempt to translate (4) into Sanskrit would result in a sentence with no adpositions whatsoever; just a verb and four NPs, one nominative, one accusative, one instrumental, and one either dative or genitive. Translating it into Russian would probably involve one preposition, collocating with the dative case.) I must confess that i do not know what within the Minimalist Program properly licenses such Cases. Best, Steven --------------------- Dr. Steven Schaufele 712 West Washington Urbana, IL 61801 217-344-8240 fcosws@prairienet.org **** O syntagmata linguarum liberemini humanarum! *** *** Nihil vestris privari nisi obicibus potestis! ***