| old_uid | 3998 |
|---|
| title | Possessives and Feature Uniqueness |
|---|
| start_date | 2008/02/01 |
|---|
| schedule | 14h-16h |
|---|
| online | no |
|---|
| location_info | salle 124 |
|---|
| summary | 1. The puzzle. Romanian, among many languages (Albanian, Latin, Czech (and other Slavic languages), Gothic and Scandinavian languages, a.o.), exhibits an intriguing alternation between adjectival and genitive-marked pronominal forms :
(1) a. b?ie?ii mei/t?i/no?tri/vo?tri
boys.the my(m.pl.)/ yoursg(m.pl.)/ our(m.pl.)/ yourpl(m.pl.)
b. fetele mele/tale/noastre/voastre
girls.the my(f. pl.)/ yoursg(f.pl.)/ our(f.pl.)/ yourpl(f.pl.)
(2) a. b?ie?ii ei/lui/lor b. fetele ei/lui/lor
boys.the she-Gen/he-Gen/they-Gen girls.the she-Gen/he-Gen/they-Gen
Quite clearly, the contrast between (1) vs (2) is an effect of the difference between pronouns 1st/2nd (both sg and pl) person pronouns and 3rd person (or ‘non-person’, in Benveniste’s terms) pronouns : the former are adjectival in the sense that they agree with the head N (expressing the Possessee), whereas the latter do not agree with the head N but instead are marked with genitive Case. Such non-uniform paradigms raise the following questions : (i) Why is it that 1st/2nd pers pronominal possessors can/must be adjectival and why is it that they cannot be marked with genitive Case? (ii) Why is it that 3rd pers pronominal possessors can/must be marked with genitive Case, and why is it that they cannot be adjectival ? (iii) Why is it the case that agreement in j-features may alternate with Genitive Case ?
2. The solution in a nutshell. Our answer will rely on the constraint stated in (3):
(3) Feature uniqueness : One and the same head cannot contain more than one type of formal features. Types of formal features : Person, Number, Gender, Case.
Formal j-features can be either ‘inherent/intrinsic’ or ‘concord’ features, which are copies of the inherent features of some other element (we will briefly compare this distinction to Chomsky’s (1995, 2000, 2001) distinctions interpretable vs. non-interpretable features and valued vs. unvalued features). The constraint in(3) implies that if a given head has an inherent Number or Gender feature it cannot inherit it via agreement from another element. Formal features, which attach to heads (e .g., Det°), must be distinguished from other features, call them ‘Index features’ (the label is inspired by, though possibly conceptually different from the index features used by Kathol 1999, Wechsler and Zlati? 2000, 2003, King et Dalrymple 2004) which may attach to either heads or full DPs. As used here, the Index feature resembles Sauerland’s (2006) highest-position j : index features are ‘interpretable’, in the sense that they convey information regarding the referent (or the antecedent of the referent). For our present purposes, the Index feature need not project its own syntactic category. The constraint in (3) allows formal features to co-occur with Index features on the same head or inside the same DP.
3. The featural make-up of pronouns. The well-known distinction between 1st/2nd pers and 3rd person pronouns (Benveniste) will be formalized by assuming that all pronouns have the Person feature (even if the value for 3rd pers is merely negative, i.e., -Person). The difference between pronouns concerns their Number and Gender features : (in the languages under discussion here) only 3rd pers pronouns have inherent Number and Gender formal features. This assumption seems very natural : (a) there are indeed no gender distinctions for pronouns other than 3rd pers (or rather non-person) ; (b) the number distinction is not marked by separate number morphemes, but rather by fused Person+Number forms (e.g., je, tu, nous, vous ‘I, you, we, you’), which may be taken to count as Persons 1,2,4 and 5 rather than 1st+sg, 2nd+sg, 1st+pl and 2nd+pl (see Wechsler (2004) for independent evidence in favor of the same assumption). This analysis of pronouns together with the constraint of Feature-Uniqueness stated in (3) explains the paradigm in (1)-(2). Because Number and Gender are intrinsic features of 3rd p pronouns, such forms cannot inherit the same type of features from the head N, and will instead be marked with Genitive Case, as shown in (3). Since pronouns of 1st, 2nd, 4th and 5th persons lack Number and Gender, they can inherit those features from the head N, yielding the adjectival forms shown in (1). One may still wonder why genitive Case cannot be assigned to those pronouns that do not carry Number and Gender features. A possible answer is that concord is obligatory whenever possible.
4. Index features. Consider the examples below :
(4) a. Tu es belle. ‘you are good-looking(sg.fem)’
b. The committee(sing) are(plur) debating.
The example in (4a) shows a case in which a gender feature appears on the predicate, although the subject pronoun lacks any formal feature for gender. The example in (4b) exhibits a contradiction between the number-marking on the subject and the number marking on the verb and on a pronominal possessor that has the same referent as the subject (Elbourne & Sauerland 1999, Sauerland 2006). In order to account for these examples, any analysis must postulate a distinction between formal features (see the 2nd pers and sg features on the subjects in (4a-b)) and Index features (see the femine and plural features, which must be postulated on the same subjects in order for the agreement features on the predicate to be accounted for). Given this distinction, the example in (4a) can be analyzed by saying that pronouns that do not carry gender features may nevertheless carry an Index feature that is valued for gender. The example in (4b) shows the possibility for formal features to co-occur with Index features that have conflicting values for Number.
Coming back to pronominal possessors, the distinction between formal and Index features is crucial for the analysis of the following Romanian examples :
(5) b?ie?ii s?i / fetele sale
boys.the SA(m.pl.) ‘his/her/*their boys’ girls.the SA(f.pl.) ‘his/her/*their girls’
We assume that the lexical entry of SA, which does not carry formal features for Number and Gender (SA is historically related to the reflexive SE, which was and still is invariable for Number and Gender), may carry an Index feature that is valued for Number. Since Index features can co-occur with formal features, the Number and Gender formal features can be copied via Agree from the head N. A similar account can be given for the agreeing 3rd p. possessives of French (sa,son,ses (sg) vs leur, leurs (pl)) and German sein [m.sg./n.sg.] and ihr [f.sg./pl.]. Note that in all these cases, the roots are different, which supports the idea that [sg] and [pl] are not FF that are projected in syntax, but are purely semantic features that constrain the Index feature.
5. Case features. How come that agreement in phi-features may alternate with Genitive Case as alternative means of syntactic legitimation ? To answer this question we will adopt Chomsky’s (2000, 2001) system, where case assignment is a by-product of Agree, and Agree is an operation of feature-valuation. In line with Pesetsky and Torrego (2001), we draw the conclusion that the structural case feature must be a feature of the case assigner. We slightly differ from Pesetsky&Torrego in assuming that the case feature is a categorial feature rather than a tense feature (because structural cases are licensed by distinct categories – T or Fin for Nominative, v* for Accusative, Poss, n or N for Genitive). Given this theory, the alternation between structural Case and agreement in formal j-features is expected, since both these means of syntactic legitimation rely on Agree (in either j-features or categorial features).
References Chomsky, N. (2000) Minimalist inquiries: the framework. In R. Martin et al. (eds) Step by step. MIT Press ; (2001) Derivation by phase. In M. Kenstowicz (ed.), Ken Hale: A Life in Language,1-52. MIT Press; Kathol, A. (1999) Agreement and the syntax-morphology interface in HPSG. In R. Levine and G. Green (eds), Studies in Current Phrase Structure Grammar, 223–274. Cambridge UPress.; King, T.H. & M. Dalrymple (2004) Determiner agreement and noun conjunction. Journal of Linguistics; Pesetsky, D. & E. Torrego (2001) T-to-C movement: causes and consequences. In Michael Kenstowicz (ed.), Ken Hale: a life in language, 355-426. MIT Press; Sauerland, U (2006). On the semantic markedness of phi-features. Proceedings of the Phi-Workshop. Wechsler (2004) Number as person. In Empirical issues in syntax and semantics 5, ed. O. Bonami & P. Cabredo-Hofherr; Wechsler, S. & L. Zlati?.(2000) A theory of agreement and its application to Serbo-Croatian. Language 76(4), 759 |
|---|
| responsibles | Dobrovie Sorin, Jouitteau, Bendjaballah |
|---|
| |