Abstract: Kayne's (1989) seminal paper on participle agreement in Italian and French has spurred numerous studies on the syntactic conditions governing agreement relations. The focus of this paper is on the semantic aspects of this phenomenon. Within the framework of Chomsky’s Minimalist Program, an analysis of French participle agreement is developed, that relates the semantic impact of overt agreement to the syntactic structure that it enforces. We argue that the interaction of Move with the projection of a morphologically strong AGR-O/v determines the position of DPs at LF and their interpretations. Modifying Diesing’s proposal for a syntax-semantic mapping, we propose that only DPs adjoined to a V-head (incorporation) can be interpreted predicatively (or non-specifically) in the sense of De Hoop (1992) while those adjoined to AGR-O/v projection receive a specific interpretation. We further take the presence of AGR-O/v in a sentence structure to be subject to Economy conditions operating on the numeration. The complex distribution of overt object agreement in French, -- optional in accusative constructions, obligatory in passives and impossible in impersonal constructions -- as well as its semantic impact on the interpretation of the triggering nominal expressions are then shown to derive straightforwardly from the interaction of Case and the Economy driven projection of AGR-O/v in the sentence structure.

Abstract: Null subject (pro-drop) Romance languages have been claimed to merge a null expletive in SpecIP in sentences with postverbal subjects. In this paper, it is claimed that economy principles should ban null expletives from entering the numeration of sentences. It is suggested that the Extended Projection Principle is covertly satisfied in these cases. Constraints against ambiguity and the Principle of Full Interpretation are satisfied under this hypothesis. The implications of this proposal for linguistic theory are discussed. The Extended Projection Principle (EPP) has traditionally been assumed to be a universal condition of well-formedness requiring that the clausal IP projection host either an argument or an expletive element in its specifier position (see Chomsky 1981,1986 and Rothstein 1983). In the Minimalist program (Chomsky 1995), the EPP is conceived as a property of the T(ense) projection and is encoded in Infl as an uninterpretable feature. I will refer to such Infl label as the [EPP] feature. Null subject Romance languages like Catalan lack phonologically overt expletives corresponding to the English it and there. Nevertheless, it is assumed that the EPP is satisfied by merging with IP a phonologically null element, 1. I would like to thank the participants of the spring 1996 CUNY Colloquium Series, to whom a previous version of this paper was presented, as well as to the audiences at the III Workshop on the Syntax of Central Romance Languages (Girona) and Going Romance 1996 (Utrecht). Thanks particularly to I. Bosque, N. Chomsky, R. Fiengo, I. Laka and G. Rigau for comments, suggestions and criticisms. Research for this paper has been supported by the grants DGICYT PB93-0893-C04 and CIRIT 1995 GR 00486. 2. The category Infl is used as a cover term to designate the functional projections dominating the predicate (see Pollock 1989; Belletti 1990 and subsequent literature). Here and throughout I abstract away from the inner articulation of Infl.away from the inner articulation of Infl. Probus 10 (1998), 219-241 0921-4471/98/010-219 © Walter de Gruyter 220 M. Carme Picallo commonly referred to as 'expletive pro\\ as shown in examples of the types (la-c) and (2): (1) a. pro va arribar ella arrived she 'She arrived.' b. pro piulava un ocell cheeped a bird Ά bird cheeped.' c. pro surt la Joana is leaving Joana 'Joana is leaving.' (2) pro es necessari que en Guillem vingui is necessary that Guillem comes 'It is necessary that Guillem comes.' Although there is no empirical evidence that a phonologically null term has entered the numeration in (1) and (2), the 'expletive pro hypothesis' has been posited for theory-internal reasons. It follows from the assumption that the [EPP] feature is strong and must be satisfied overtly. In the discussion that follows, I examine this assumption by proposing the hypothesis that the [EPP] feature is weak in Catalan and, possibly, in most null subject Romance languages. This hypothesis has a number of consequences because it affects the syntactic 'architecture' of many constructions. It also makes a series of predictions concerning word order and covert movement operations. In this paper, I will explore only some of these consequences and limit the discussion to declarative sentences. This paper is organized as follows: Section 1 is devoted to analize the conflict that arises between economy principles and the EPP in sentences of the types (la-c). I suggest that an economy principle proposed in Chomsky (1995) bans the selection of null expletives, hence, economy appears to override the EPP, as has traditionally been assumed. In Section 2, monoargumental constructions and subject DP preposing are discussed. Section 3 analyzes transitive structures and constraints on word order. Section 4 is devoted to discussing /r-extraposition constructions. A brief summary and conclusions follow in Section 5. 1. Expletive pro^^ and economy principles. An analysis of the properties of the expletive item in the representation of sentences of the types (1) shows that a tension arises between economy princi3. See, however, Silva-Villar (1996) for data showing that some pro-drop Northwestern Iberian varieties appear to have phonologically realized expletives. On the Extended Projection Principle and null expletive subjects 221 pies and the EPP. In this section, it is suggested that economy and constraints against ambiguity should ban null expletives from entering the numeration of sentences. Except for the absence of phonological features, an expletive pro associated with a DP, as in (la-c), is assumed to have the properties of the English expletive there (see Chomsky 1995: 274). Such null expletive will be referred to as prothere to facilitate the discussion. This term enters the numeration with the categorial feature D, but it is devoid of Case and inherent Phi content. Under the assumption that the [EPP] feature is strong in the null subject Romance languages, /?rothere merging with IP allows the EPP to be overtly satisfied. The weak [Nominative] and [Phi] features of Infl attract (and are covertly checked) by the formal features of the postverbal DP (FFsDP), which agrees with the verb in standard Catalan. The checking procedures are abstractly represented in (3): (3) [IP P^there([EPP]) Cl-max ^DP([Case],[Phi])· · Ή· · -DP...]] The hypothesis that a pro^eK is a term in constructions of the types (la-c) appears to be in conflict with the economy principle (4) suggested in Chomsky (1995: 294): (4) Alpha enters the numeration only if it has an effect on output. According to (4), a term or a feature cannot be selected if it does not provide some instruction to the phonological or to the interpretive component. Principle (4) should ban prothere from being selected because this element has no PF or LF effects. It is phonologically null and does not trigger overt movement operations. In addition, it is not a legitimate LF object: it is not an argument and its formal content does not contribute to the interpretation of the sentence. Prothere is conceived as consisting of an uninterpretable categorial feature that is introduced to satisfy computational needs exclusively. The expletive feature checks [EPP] in Infl and must, in turn, be covertly checked by the interpretable categorial feature of its associated DP argument if the derivation is to converge (Chomsky 1995, 364). The procedure is abstractly represented in (5): Alexiadou and Anagnostopoulou (1995, 1996) have also proposed that Spanish and Greek lack expletives. As opposed to Belletti (1988) and Lasnik (1995), among several others, Chomsky (1995) does not assume that the argument associated with a there-like expletive receives (inherent or structural) Partitive Case. 222 M. Carme Picallo

Abstract: The goal ofthis paper is to present an analysis ofaccusative clitic doubling that can account for some of its traditional problems and also for a new problem related to its aspectual properties. The analysis I will be proposing is based on the assumption that the syntax of direct arguments crucially affects the aspect of a predicate since it crucially affects what is atAgrO by the time the terminativity of the VP is calculated.

Abstract: This article treats the fairly well known process of syllable-final consonant loss in Old French from an Optimality Theoretic perspective. I show that the Old French NoCoda effects are best seen as derived from the interaction of a constraint militating in favor ofbimoraic syllables, a sonority-based constraint on moraicityy and PARSEFeatures. The OT account developed here allows straightforwardly for the grouping together of the sonorant consonants and /S/, as well as for the maintenance of certain features in two of the specific changes. This account is preferable to one in terms ofOT's NoCoda constraint in that it is able to account for the compensatory lengthening which accompanied the change, as well as the fact that the change stopped short of M in many dialects.

Abstract: This paper addresses negative ne, N eg0, in French which is restricted to contexts of sentential (as opposed to constituent) negation. This association is attributed to a single mechanism: ne is licensed and sentential negation is marked via Dynamic Agreement in an S-Structure spec-head configuration with a negative operator in SpecNegP, whereby the [+NEG] feature of the operator is transmitted to Neg°. This conclusion explains central propeties of the negative operator pas. While pas can appear in situ, below SpecNegP, it cannot license nefrom such a position and it has narrow scope; to license ne and have wide scope, pas must overtly raise to SpecNegP. This analysis works well where sentential negation is marked by pas but there are problems where sentential negation is marked, for example, by plus or rien, which will be termed associates. In such contexts, ne can be licensed and sentential negation marked without the associate occupying SpecNegP at SStructure. Other authors have solved the problem by assuming: (a) that associates are negative operators like pas; (b) that they raise to SpecNegP at LF; and, (c) that it is at LF that ne is licensed and sentential negation marked. This analysis is problematic since it provides no principled account for why pas cannot raise to SpecNegP at LF too. In this paper, in contrast, S-Structure is maintained as the relevant level of representation. The inability of pas to mark sentential negation and license ne from below SpecNegP is thus explained. As for the problematic associates, it is argued that these are not in fact negative operators. Rather, the negative operator which occupies SpecNegP — licensing ne and marking sentential negation — is a non-oven element: Op, an alternative surface form (with distinct licensing conditions) to pas of the same lexical item. Crucially, Op occupies SpecNegP at S-Structure and the original analysis ofnt licensing can be maintained.

Abstract: The issue of the formation of finite sentences in which lexical verbs adopt the imperative morphology is addressed taking recent contributions to the topic (Rivero 1994, 1997; Rivera and Terzi 1995; Zanuttini 1991, 1994; Hulk 1996; Rooryck 1992) as a point of departure. The role of Greed and feature checking (Chomsky 1995) deriving imperative sentences in Old Spanish and within the Modern Romance domain in general is revisited, reducing the role of the former as a theoretical principle regulating the distribution between class I and class II languages (Rivero and Terzi 1995). The split between true and suppletive imperatives is refined arriving at a new distribution that favors abstract second-person singular morphology which, interacting with negation, parametrizes Romance languages in two different groups covering Southern and Northern Latin areas. In memory of Joan Corominas (1905-1997)

Abstract: The principle of compositionality is essential to account for the ability to understand the meaning of novel utterances on first hearing. Intersective adjectives fall naturally in this picture, but subsective adjectives and non-predicative adjectives appear to hold a much more complex relation with the noun. This has brought researchers to weaken compositionality and isomorphy with type shifting rules or multiple semantic entries for some adjectives. Kamp andPartee (1995) propose a partial solution: some subsective adjectives would actually be intersective, but context-dependent. Evidence from French suggests a more general solution. French allows both pre-N and post-N adjectives in the NP, but whereas a post-N adjective combines with the head N as a fully closed-offfunctor category, a pre-N adjective modifies a component internal to N. The strongest version of compositionality and isomorphy is maintained: the same adjective in pre-N and post-N position appears to have different meanings because it composes with different components of N. In comparison, derivational analyses of adjective placement require disconnected stipulations that conspire to derive the effects that one gets from the surface structure that the Integral Merging analysis arrives at in a much more parsimonious manner. Moreover, current derivational analyses of adjective placement do not address the core of the issue of compositionality.