Nereus I: Semantic and syntactic aspects of specificity in Romance languages

International workshop at the Universität Konstanz, October 2002

Organizer

Description

The semantic-pragmatic category specificity is a very recent category – it was introduced in the late sixties. Since then it has been used in many descriptive grammars for describing a great variety of grammatical phenomena, such as different articles, case markings (e.g., differentiated object marking), the modality in relative clauses, word order and information structure, clitic doubling, to list only a few. Thus specificity seems to be a category that is realized in very different grammatical areas.

The workshop intends to investigate the semantic and syntactic aspects of this category in the Romance language family. Romance languages provide an interesting field for investigating specificity since they show some common realizations of specificity (such as the interaction of modality in the relative clause with the specificity of the head noun), but they also show language particular grammatical reflexes of specificity, such as Spanish a or Romanian pe.

The workshop brings together experts in Romance languages that will discuss different semantic and syntactic aspects of specificity. We hope that the presentations and discussions will provide us with a more detailed picture of the linguistic realization of specificity in Romance languages, but also about a more detailed linguistic model of specificity in general.

Invited Speakers:

Specific Indefinites in Romanian and the Marker PE

Agnes Bende-Farkas, University of Stuttgart, Germany

Romanian has an Accusative preposition, pe, which is usually said to mark definiteness, or specificity (Farkas 78). Pe is obligatory with proper names, (strong) pronouns, overt definites, and certain quantifiers (like the universal quantifier toti, toate `all'). Pe is optional with indefinites; when pe does occur with them, it is said to mark specificity (Dobrovie-Sorin 94). (1b) can only have a specific construal, whereas (1a) can, but need not have one.

• (1) a. Ion iubeste o actrita.
• I. loves one actress. "John loves an actress.''
• b. Ion Ion iubeste pe o actrita.
• I. loves PE one actress. "John loves a certain actress.''

This paper concentrates on semantic and pragmatic issues related to pe-marked indefinites. Some new data are presented, with the hope that they may shed light on currently debated issues in the literature on indefinites --- or at least one can learn how these problems manifest themselves in Romanian (as opposed, say, to English).

The first task of the paper is to investigate the `specificity' properties of pe-marked indefinites: roughly, whether pe has to do with scope, with partitivity or with the indefinite's picking out a particular individual. The second task is to uncover possible grammatical or lexical constraints on pe. This paper presents two such constraints, notably, with certain verb classes and with Dative reflexive clitics. The relevance of this point is architectural, so to speak: In such `environments' pe is not possible, even if context clearly warrants a specific use for the indefinite. There is thus a `pragmatic residue' with specificity: the term `residue' is meant as a reminder that not all instances of specific use can be encoded by grammatical or morphological means. These constraints also show that pe-marking is not purely semantics- or pragmatics-driven. Pe being part of a grammatical inventory, its licensing depends on other grammatical factors as well.

Overridden Specificity

Susann Fischer, University of Stuttgart, Germany

We know that Modern Romance clitics notoriously fill syntactic positions which are not available to other DPs and vice versa. See the following examples from Modern Catalan. For instance, (1c) shows that displacement of a lexical DP to a position between the subject and the finite verb is impossible, whereas it is allowed for clitics (1b). On the other side clitics are not allowed to stay within the VP (1d).

• (1) a. La Mercè veu en Joan.
• b. La Mercè el veu
• c. *La Mercè en Joan veu
• d. *La Mercè veu lo

What appears to be syntactically encoded in these cliticization structures is the ”familiar” interpretation of these constituents, whose referent has to be identified with entities which have already been introduced in the discourse or are physically salient. In the literature, clitic movement has often been related to ”specificity effects” (cf. Sportiche 1996, Uriagereka 1995, Corver and Delfitto 1999; and Martins 1995a, 1995b). Here specificity is clearly related to the definition of ”definiteness” in terms of the identity relation, found in Enç (1991), and also clearly overlaps with the use of ”familiar” indexes in a DRT-related approach to clitic movement, according to which pronouns move to VP-external functional projections, in order to avoid to be interpreted in the scope of existential closure. In Old Romance, however, the pattern is not as clear-cut. Clitics are allowed to precede and follow the verb in matrix as well as in embedded sentences. See the following examples from Old Catalan (2).

• (2) a. E lo cavaller li respòs e dix que ... (Llull/41)
• b. E ab la doctrina qui li donà son pare, (Llull/4)
• c. e donà-la per muler a l’emperador de Castela; (Desclot/1)
• d. E Déus donava-ls manera per què la caritat e devociò moltiplicava. (Llull/188)
• e. ab los crestians que volien-los sobtar, (Desclot/122)

(2c) is usually interpreted as a last resort movement of the verb in order to protect the clitic to appear in initial position, also known as the Tobler-Mussafia effect, i.e., a purely phonological requirement of Old Romance clitics that need a host to their left. However, this cannot be taken as an satisfactory explanation for (2d) where the verb does not need to protect the clitic, since the clitic is already protected by the subject or the relative pronoun that precedes it. Furthermore, in Old Catalan objects are allowed to appear between the subject and the finite verb (3b).

• (3) a. lo hermità hac satisfet al caveller de la demanda que li hac feta, (Llull/37)
• b. La bona dona a son fill ajudar no podia e per la gran dolor que havia, (Llull/355)

How do we have to interpret the post-verbal clitics in Old Catalan with respect to specificity? Why are full object NPs not allowed to intervene between subject and verb in Modern Catalan? Has there been a change in how specificity is encoded in Old Catalan compared to Modern Catalan?

This paper will argue that there is a difference with respect to specificity in Old Catalan compared to Modern Catalan that is connected to the development of the determiner system from Proto-Romance to Modern Catalan. As long as the determiner system has not been fully grammaticalized the specificity of the clitics could be overriden by the clitics’ being in the scope of an operator that changed the sentence interpretation. More precisely, I will argue that in Old Catalan word order contributes to textual information organization (cf. Fischer 2000), whereas in Modern Catalan a lot of the discourse organisation is done by the use of definite vs. indefinite determiners. As long as the determiner system was not fully developed other means were needed in order to structure text discourse.

Prominence Scales and Unmarked Word Order in Spanish

Rodrigo Gutierrez, UC Santa Cruz, USA

This paper investigates a number of word order phenomena attested in Spanish, concentrating on the unmarked word order of this language. The core proposal developed here is that the unmarked word order in Spanish is not the result of some licensing condition related to the subject (i.e. Case, agreement, etc.), but rather results from the interaction between the Extended Projection Principle (EPP) and considerations regarding structural markedness.

The analysis developed here argues that, in the unmarked case, the EPP in Spanish is satisfied in the specifier of the highest inflectional projection by the argument of the verb that ranks highest in the thematic hierarchy, which may but need not correspond to the grammatical subject. To disassociate the constituent that satisfies the EPP from any specific grammatical relation, I propose that it be referred to as the Pole of the clause. What determines then the unmarked word order of clauses with different classes of predicates is the sensitivity of the Pole position to the semantic role of the constituent that occupies it.

I propose that Spanish clauses with different constituents in the Pole position have different degrees of structural markedness, depending of their semantic role. Agents and experiencers constitute the least marked instance of a Pole. This explains why the preverbal position in this language is occupied by a subject agent, whenever there is one, or by a dative experiencer, in the absence of an agent. Other arguments and adjuncts which rank lower in the Thematic Hierarchy constitute more marked instances of a Pole. I argue that beyond a certain degree of structural markedness (when the constituent that would satisfy the EPP ranks low in the Thematic Hierarchy) it is better not to satisfy the EPP altogether. This explains a number of Spanish verb-initial constructions where the highest inflectional specifier is left empty. I suggest that Optimality Theory (Prince & Smolensky 1993), where well-formedness conditions like the EPP are interpreted as violable constraints, provides a straightforward analysis of this state of affairs. Finally, I provide some discussion as to whether the Pole position in Spanish is sensitive to other semantic prominence scales besides the Thematic Hierarchy.

Animacy, Specificity, and Definiteness in Spanish - A Research Project

Klaus von Heusinger, University of Stuttgart, Germany & Georg A. Kaiser, University of Constance, Germany

We like to present some preliminary ideas to a potential research project on the syntax-semantics-interface with respect to the grammatical realization of animacy, specificity and definiteness. Animacy is a lexical feature of NPs that assigns a property to the referent, specificity expresses a semantic property that determines the referent in a particular way, and definiteness is a discourse-pragmatic property of the discourse item representing the NP. Even though these three categories are associated with quite different semantic domains, their grammatical reflexes are closely interrelated, as it can be shown for several phenomena in Spanish. The project intends to investigates the semantic nature of the categories, their interaction in terms of hierarchies, their syntactic encoding, and finally tries to develop a uniform description of otherwise unrelated phenomena.

"Possession", Noun Combining, and Specificity

Daniel Jacob, University of Cologne, Germany

This paper starts from the assumption that so-called 'possessive' constructions (i.e. NPs whose modifier is a possessive pronoun or a genitive NP) have to be considered a special syntactic device for the syntagmatic combining of two nominal concepts, where one nominal term (the genitive modifier) serves to indicate a domain for the reference expressed by the other nominal term (the head noun), e.g. Peter's first total eclipse: "the first total eclipse within the universe defined by "Peter". From this perspective, the genitive construction is compared to other NP constructions combining two or more noun concepts within a NP: nominal composition (German: die Kongressakten) and so called "relational adjectives" (French: la voiture présidentielle), with respect to the different reference types possible for the modifier position. It seems that the specificity of the modifier concept is a crucial criterion for the selection of one device or the other (French la voiture présidentielle vs. ? la voiture chiraquienne; German ? die Konstanz-Akten vs. die Konstanzer Akten).

However, these rules seem to have an approximative character, allowing less prototypical use of these devices with specific modifier concepts: German das väterliche Haus, die Stasi-Akten, French le bureau chiraquien.This use is particularly current in stylistically marked contexts (ironic speech, newspaper-style ...).

Specificity Marking and Specificity Types in Spanish

Manuel Leonetti, Universidad de Alcalá, Spain

It has been often pointed out that the term specificity covers a number of different notions (scopal distinctions, partitivity, epistemic specificity, referentiality; see Farkas 2001 and von Heusinger 2001) and is thus usually employed in a rather un-specific way. Different hypothesis have been proposed to give an account of the basic specific / non specific distinction for indefinites, from the classical scope theories to pragmatic and presuppositional ones. Bearing in mind these difficulties, I intend to review the mechanisms for specificity marking in Spanish (determiners, mood in relative clauses, object marking, syntactic position, prenominal adjectives…); my aim will be that of determining what kind of specificity is associated to each mechanism and what is the particular contribution that such mechanisms make to utterance interpretation. My two basic assumptions are the Relevance-theoretic distinction between linguistic semantics and explicatures(s) of an utterance, and the idea that topicality and information structure can be related to the use of a significant number of specificity markers, not only in Spanish but also in a variety of natural languages. This is consistent with a view of specificity as an epiphenomenon, i.e. a pragmatically inferred part of the propositional explicature.

Clitic doubling and clitic-left dislocation in Spanish and Greek as native and as L2 grammars

Teresa Parodi, Cambridge, UK

The presentation deals with the status of clitic doubling and clitic-left dislocation in Spanish and Greek both in native grammars and in that of English learners of these languages.

Based on syntactic and morphological evidence object clitics in Spanish and Greek are analysed as agreement markers along the lines of Suñer (1988) and Franco (2000), among others. Features of definiteness, specificity and case rule the cooccurrence of clitics with a coreferent NP in clitic doubling structures. Furthermore in clitic-left dislocation clitics in Spanish and Greek license the long-distance relation (Cinque 1991). English, on the other hand, has no clitics and uses strong pronouns (or empty categories) instead. Due to the different nature of clitics and pronouns there is no one-to-one correspondence between them, which gives rise to interesting acquisition problems.

The conditions on the cooccurrence of a clitic and a coreferent NP in clitic doubling and clitic-left dislocation have been tested on native speakers and on learners of Spanish and Greek. The results will be compared to those of a similar test carried out on Macedonian, a language close to Spanish and Greek with respect to clitic doubling.

Under the acquisitional perspective the paper aims to discuss morphological development as well as the issue of optionality in the acquisition of a language morphologically richer than the L1.

• Cinque, G. (1991): Types of A’-dependencies. Cambridge, Mass.: MIT Press.
• Franco, J. (2000): Agreement as a continuum. The case of Spanish pronominal clitics. In Beukema, F. & M. den Dikken, Clitic Phenomena in European Languages. Amsterdam: Benjamins. 147-189.ess.
• Suñer, Margarita (1988): The role of agreement in clitic-doubled constructions. Natural Language and Linguistic Theory 6, 391-434ess.

Specifically Brazilian

Uli Reich, University Cologne, Germany

Since the target of our workshop is the comparison of Romance languages with respect to the grammatical means expressing the logical and cognitive category specificity, my talk tries to contribute to it by introducing facts from Brazilian Portuguese (BP) into the discussion.

The expression of specificity can be easily divided into (i) explicit lexical forms like quantifiers, (ii) explicit morphosyntactic forms like modality in relative clauses (que tenha/tem cabelos azuis) and (iii) implicit forms, in which the specification of NPs seems to be rather a secundary effect of semantic properties of other domains (cf. the difference of the reading of the object NP triggered by the imperfect/perfect tense of the verb, e.g. Meu amigo queria/quis beijar uma alemã, or by the semantics of the verb as discussed by Lauri Karttunen 1976). Differences between languages are more likely to occur in the last two realms of means than in the first. Of particular interest in BP is the choice of either zero anaphora or free pronouns to corefer with specific or nonspecific discourse topics, an aspect of this language upon which Brazilian linguists have recently paid much attention (cf. the work of Sonia Cyrino and Mary Kato). The specificity of specificity in BP is also observable in other domains of its grammar, like the use of zero articles for specific singular NPs (Lhe desejo ótima viagem!) and other interesting phenomena. The Romance perspective will be opened by a systematic comparison of the Brazilian facts with those of the closest relatives of BP: European Portuguese and Spanish.

Indefiniteness, countability and specificity in Old Italian

Elisabeth Stark, University of Munich, Germany

Logical, pragmatic or discourse-representation-based approaches to (in)definiteness describe its basic value either as the opposition of ‘inclusiveness/exclusiveness’, following the ‘uniqueness approach’ of Russell (cf. Hawkins 1978) or as ‘identifiability’, following the ‘familiarity approach’ of Christophersen 1939 (cf. Karttunen 1976, Heim 1988, Kamp/Reyle 1993; cf. Lyons 1999 chapters 1, 4 and especially 7). The apparent double nature of nominal (in)definiteness comes from the different semantic values of the articles as prototypical markers of nominal (in)definiteness in different syntactic environments: Whereas in predicative position, the opposition between the definite and the so-called indefinite article seems intuitively to express ‘exclusiveness’ vs. ‘inclusiveness’ (John is a victim of Paul vs. John is the victim of Paul, cf. Declerck 1986, 29), it seems to express ‘(non)-familiarity’ or ‘(non-)identifiability’ of the intended referent by the hearer in argument position (I will buy a car this morning vs. I will buy the car this morning, cf. Lyons 1999, 2). This latter opposition comprises yet another one, often called the&Mac226;specific - non-specific distinction’ (more precisely ‘epistemic (non-) specificity’: the intended referent of a car can be known (‘a certain car’, specific) or unknown to the speaker (‘whatever car’, non-specific)).

While there being other definitions and (semantic) distinctions referred to under the heading of ‘(non-)specificity’ (particularly important: ”scopal specificity” in opaque contexts, for a survey cf. von Heusinger 1997, 126-134, see also Farkas (to appear), von Heusinger (to appear)), the basic distinction between ‘one particular entity’ and ‘any arbitrary menber of the class’ (cf. Lyons 1999, 165) is to be considered as fundamental when it comes to the emergence of systems of nominal determiners, e.g. in Romance languages in the Middle Ages. Noun phrases with specific interpretation are more likely to be explicitely determined than non-specific ones (cf. Blazer 1979 for Old French or Selig 1992 for late Latin, Stark (to appear) for Old Italian).

Messere Jacopino Rangoni, nobile cavaliere di Lombardia, stando uno giorno a una tavola, avea due ingaistare di finissimo vino innanzi, bianco e vermiglio. Un giucolare stava a quella tavola, e non s’ardiva di chiedere di quel vino. Avendone grandissima voglia, levossi suso e rese uno muiolo e lavollo smisuratamente bene e davantaggio e, poi che l’ebbe così lavato e sciacquato molto, girò la mano e disse: ”Messere, io lavato l’hoe”. E messere Jacopino diè della mano nella guastade e disse: ”E tu il pettinerai altrove che non qui”. Il guillare si rimase così, e non ebbe del vino. (Novellino, 43th novella)

In the domain of nominal indefiniteness, however, not every noun or noun phrase with specific interpretation can be determined by the indefinite article derived from the numeral ‘one’ (uno in Italian). The different occurrencies of (the very specific!) vino in the above example show that it can easily remain undetermined (finissimo vino) or be marked by a special determiner for ‘mass interpretation’, the partitive del (being grammaticalized in the main Romance languages only in French and Italian). This leads to the hypothesis that in a determined stage of determiner grammaticalization in Romance, explicit indefinite determination of nouns is to be related to the dimension of apprehension (cf. Seiler 1978, Lehmann 1991), i.e. classification of the noun’s denotation as ”an undifferentiated concept or as an individual” (Lehmann 1991, 206). We will show, based on evidence from Old Tuscan texts from the 14th to the 16th century, that the main cause of indefinite determiner grammaticalization in Old Italian is the need of classification in the domain of indefiniteness (uno vs. del), being related to the original value of (in)definiteness, i.e. the quantitative one (‘inclusiveness’ vs.’exclusiveness’ or partitivity), while the textual value (‘familiarity’) develops only later (as an implicature), and is typical only for fully grammaticalized determiner systems (cf. Leiss 2000). Specificitiy in this prospective remains still the key concept for an adequate description of the distribution of determiners in Old (Italian) texts, but can be described as related to the classification of concepts as either ‘countable’ or ‘massive’ (‘specificity as classifying partitivity’), only acquiring later the textual dimension of ‘giveness’ or ‘anchoring’ (‘textual partitivity’, cf. Enç 1991, von Heusinger (to appear)).