LECTURETTE #17: FURTHER ON CASE-ASSIGNMENT There are a few questions about Case Theory that i feel i need to try to answer, or at least discuss, in this Lecturette. 1. HOW MANY CASES ARE THERE? When talking about overt, morpho-phonologically marked case in traditio- nal grammar this is a relatively easy question. (Or at least it is for Indo-European languages. As a part-time student of Uralic languages, i find that in many members of that family there is a sort of gray area where one is not sure whether one is talking about case-affixes or about other sorts of 'Fillmorean' case-markers, e.g. adpositions. Which means of course that there's controversy as to exactly how many 'cases' there are in, e.g., Finnish or Hungarian.) But in PPA's Case Theory, as i ex- plained in the last Lecturette, we're talking about *abstract* Case. So there's no a priori reason why there should be any connection. Be that as it may, one often encounters the assumption (often tacit) in the literature that the set of abstract Case feature-values in a given language is isomorphic with the set of overt morphological cases in its grammar. In English (still the most discussed language in the PPA litera- ture) there are three morphological cases, at least in the sense that our personal pronouns exhibit a maximal 3-way case distinction: i/me/my, he/him/his (forms like 'mine' and 'hers' are typically treated as non-cli- tic variants of the 'genitive'/'possessive' forms). (Dick Hudson has a paper arguing that this superficial distinction in English is a phantom and that English has no morphological case at all. He has a point, but i'm not going to go into that can of worms right now. For the moment, let's just accept the assumption that English personal pronouns exhibit a 3-way case distinction.) And typically in the PPA literature one finds the assumption (often tacit, or at least unargued) that at the abstract syntactic level English also has three Cases. There's the Case for the subject of a finite clause, which we can call 'nominative' or 'subjec- tive'. There's the 'genitive' or 'possessive' Case exemplified by the first NP in each of the phrases in (1). And there's the Case for com- plements of verbs and prepositions, which we can call 'accusative' or 'objective'. There doesn't seem to be any reason to make finer distinc- tions within that category; the NP 'you' is regarded as having the same Case in all five of the phrases in (2), even though in each it has a dis- tinct theta-role, a distinct Fillmorean Case-relation, and in some lan- guages (e.g., German or Sanskrit or Hungarian) its equivalent would bear a distinct morphological case in each. (1) a. your cat b. Sam's cat Bellhop c. Bellhop's feeding by Sam d. Sam's feeding of Bellhop (2) a. love you b. remember you c. to you d. with you e. think of you For the moment, we're going to say that (i) English grammar evidences a maximal 3-way distinction with regard to morphological case, ergo (ii) English grammar also recognizes a 3-way distinction in abstract, syntac- tic Case, and leave it at that. And we're similarly going to assume that languages with richer case morphology than English, e.g. Latin, Sanskrit, or, God help us, Hungarian, have as many abstract syntactic Case distinc- tions as they have overt morphological ones. But what about languages like Chinese, which reputedly have no overt morphological case distinc- tions at all? There are, to the best of my knowledge, two answers to this question, which may not be mutually exclusive. One is that the 3-way distinction evidenced in English is the minimum allowed by Universal Grammar, so Chinese, etc., must have a 3-way Case distinction too, even though there is little or no evidence for it superficially. The other is that our original question, How many Cases are there?, isn't terribly important. Like theta-roles, what's important is how NPs get Case, not which Case they get or how many are available in the grammar. I don't find either of these answers terribly satisfying, but for now i'm going to move on to this more 'core' issue. 2. WHAT CAN ASSIGN CASE? First of all, as mentioned in the previous Lecturette, verbs and adpositions, i.e., -N lexical heads, can assign Case. That is, some verbs have the power to assign Case to their complements; not all verbs do. As also noted in Lecturette #16, passive verbs are assumed to have lost at least some of their Case-assigning power, with the result that their base-generated complements (which receive the theta-roles these verbs still have in their gift) must move elsewhere to get Case and pass the Case Filter. And many verbs, e.g., copulas, may have no Case-assign- ing power at all. This issue is related to, though not identical with, the traditional distinction between transitive and intransitive verbs. And, as i have mentioned previously, adpositions may be considered to be merely 'transitive adverbs', i.e., adverbs that take NP complements and have the power to assign Case to them. So again, we get a connection between Case-assigning power and transitivity of a sort. (I'm going to risk some tangential remarks here about the way Case Theory can be involved in some ambiguous or 'garden-path' constructions, especi- ally in certain registers. The fact that adpositions are merely 'transi- tive adverbs' is further complicated by the fact that some adverbs, like some verbs, seem to be optionally transitive. In the past few months we have twice sung in church the following early 18th-century lyric by Philip Doddridge: (3) The nations round Thy form shall view, With luster new Divinely crowned. The first few times i came across this text, i kept trying to read it as (3') [The nations [round thy form](PP)](NP) shall view ... but 'view', at least for me, is an obligatorily transitive verb, and this reading didn't provide a direct object NP ('lustre', the only remaining nominal, is clearly the complement of the unredoubted preposition 'with'). Eventually, it dawned on me that 'round' isn't a preposition in this sen- tence, even though it's immediately followed by an NP that is interpreta- ble as its complement, but a simple adverb or adjective modifying (more or less) 'the nations', and 'thy form' is the object of the verb, as in (3'').) (3'') [The nations round](NP) [thy form](i) shall view [t(i) with ...](NP) Is there anything else that is empowered to assign Case? Well, Alec Marantz in his dissertation (published in 1984 by MIT Press as Linguistic Inquiry monograph #10 under the title On the Nature of Grammatical Rela- tions) suggested (p. 21 of the published version) that, in languages that have them, morphological case-markers could assign abstract Case as well, and as far as i know this proposal has gone for the most part uncontested (there may be some grumbling about it here and there, but i don't know of anybody who has presented an organized, reasoned-out case against it); Chomsky seems to have bought it (Knowledge of Language, p. 215). I will have more to say about this in the next Lecturette. It has already been hinted that a functional head such as Infl or Agr, whatever one wants to call it, can assign Case, at least in that it is assumed that the subject of a finite clause gets its Case from the func- tional head of the clause. More generally, there is a common assumption that Case assignment is of two types, Structural and Inherent. Structural Case is assigned to an NP purely on the basis of its overt con- stituent-structural position, and depends solely on the presence of an ap- propriate governor. Inherent Case requires in addition that the governor assign a theta-role to the NP in question. This is not a bidirectional distinction; while the assignment of a theta-role is a necessary condi- tion for the assignment of Inherent Case, Inherent Case is not a necessa- ry consequence of theta-role-assignment. Thus, a great many verbs assign theta-roles, but there is no reason to suppose that the complements of most of them get Inherent Case instead of Structural Case. Chomsky (Know- ledge of Language, pp. 186-204) argues that nouns in English have the pow- er to assign Inherent Case, and that the 'of' in phrases such as those in (4) is merely the manifestation of this Case, which he calls 'genitive'. As Haegeman points out (Introduction to Government & Binding Theory, p. 165), this hypothesis involves a 'mismatch' between the Inherent abstract 'genitive' Case assigned by the head noun and the Structural 'objective' Case assigned by the preposition that realizes it. I have other problems with it, in that it complicates the description of Case-assignment in the Spec-NP position. As noted earlier, NPs in the specifier position of other NPs typically get overt 'genitive' case, while NPs in complement position of other NPs are embedded in prepositional phrases within which they presumably get 'objective' case. These cases are clearly not the same, nor are they interchangeable (though as can be seen from (1c-d) above, their theta-roles may be), as is clear from (5). (4) a. the colour of the car b. the painting of the car (5) a. your treatment of me b. my treatment of you c. *me treatment of your d. *you treatment of my The distinction between Structural and Inherent Case is more relevant in languages that combine rich morphological case-marking systems with 'quirky' cae, i.e., in which there are words belonging to a Case-assigning lexical class (typically verbs) whose Case-assignment differs from the default for that class. Many of the Indo-European languages, ancient and modern, fall into this category. Thus, in German most transitive verbs (here defined as verbs taking NP objects, i.e. which have theta-roles to assign to complements) assign accusative case to their objects. But there are exceptions. For instance, 'helfen' 'help' assigns dative case to its object, while 'denken' 'remember, think about' assigns genitive. Crucially, the complements of these verbs retain these 'quirky' cases (and, incidentally, fail to control verb-agreement) even when the verbs are passivized; cf. (6). This is typical of many Indo-European languages. (6) a. Sie half IHM. (lit.) 'She helped him.' b. IHM wurde geholfen. (lit.) 'Him was helped.' c. Ich dachte DEINER. (lit.) 'I remembered your.' d. DEINER wurde gedacht. (lit.) 'Your was thought of.' 3. UNDER WHAT CIRCUMSTANCES CAN CASE BE ASSIGNED? There is general agreement that Case is assigned under government: an NP must receive Case from a head that governs it, bearing in mind minimality conditions and barriers as explained in Lecturette #15. It has also been suggested that adjaceny is a further condition for Case-assignment, in that in English a complement must be adjacent to its Case-assigning gover- nor; cf. (7). However, it is evident from (8) that such an adjacency con- dition doesn't operate in German or French. (7) a. Terry often drinks beer. b. *Terry drinks often beer (8) a. ... dass Hermann das Bier oft trinkt. b. Pierre boit souvent la bi`ere. The availability of these options in German and French is presumably due to movement operations that are characteristic of those languages but not of English. The German clause in (8a) results from the movement of the di- rect object out of the VP and its adjunction to the maximal projection of some intermediate functional node (i.e., a node whose maximal projection dominates the adverb 'oft' but not the subject 'Hermann'). The French clause in (8b) results from the movement of the verb out of VP to Agr. Thus, these two clauses are properly analyzed as in (8'), in which in both cases the base positions of object and verb are adjacent. Bearing in mind that Case and theta-roles are properly understood as being as- signed not to NPs per se but to chains, the fact that a trace of the ob- ject is adjacent to the verb in (8a) is just as good as having the object itself in that position, for purposes of Case-Assignment. And it is gene- rally understood that a moved verb can still assign Case through its trace. (8') a. dass Hermann [das Bier](i) [oft t(i) trinkt](VP) b. Pierre boit(i) [souvent t(i) la bi`ere](VP) Such movement is not characteristic of English. In Modern English, auxi- liaries like 'be', 'have', and 'can' are the only verbs that normally ma- nifest in the position of the functional head of S; other verbs, crucial- ly the ones with real theat-grids and Case-assigning power, remain within VP throughout the derivation (I'm here reporting the common analysis with- in PPA; as many of you are aware, i am myself sceptical about the whole notion of functional heads and of movement to them, and am currently work- ing on a critical examination of that whole theoretical and empirical question); sentences like those in (9), while not uncommon in Early Modern (i.e. 16-17th-century) English, are marginal in the 20th century, gramma- tical (probably because of our collective historical consciousness, resul- ting from an extensive literary history buttressed by the educational sys- tem) but seldom used. And while it is possible in English for object NPs to move out of the VP, in general this only happens if they are either topicalized as in (10) or sufficiently 'heavy'. What is meant by 'heavy' in 'Heavy NP Shift' has always been subject to a certain amount of debate and inconsistency amongst linguists, but it is generally understood in a vague way that the more words, syllables, or phonemes make up an NP the 'heavier' that NP is and the more likely it therefore is to be shunted to the end of the clause. Thus, while (11a) would be regarded at the very least as extremely awkward (11b) is quite unremarkable, even though from a purely syntactic point of view they are structurally identical. Thus, in Modern English there is much less opportunity than in many other lan- guages for verbs and their objects not to be adjacent to each other (leav- ing aside, of course, the question of double-object constructions such as (12), in which the verb has two objects and, by the conventions of English syntax, both must appear on the same side of the verb, necessitating that one fail to be adjacent to it in terms of surface order.) (9) a. She knows naught that knows(i) not t(i) this. b. The dog all this while sheds(i) not t(i) a tear. c. Begin(i) not t(i) your first building upon the ruins of the innocent. d. I care(i) not t(i) a straw for you. e. He discovers(i) not t(i) his whole design at once. f. Ask(i) not t(i) what your country can do for you. (10) Peas(i) i like t(i). (11) a. *?Morgan tossed across the river it. b. Morgan tossed across the river the knobby green branch that Sam had found earlier. (12) Terry gave the baby a toy. This Lecturette has gone on long enough, and i'm afraid i'll have to put off to the next one my planned discussion of the emendations of Case Theo- ry characteristic of the Minimalist Program. 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! ***