Configurationality and the Direct Object Clitic in Bulgarian Veronica A. Gerassimova T. Florian Jaeger Linguistics Department, Stanford University
Introduction ¾Configurationality is a basic assumption in some rulebased theories, especially those which employ the idea of a Universal Grammar (e.g. GB, MP theories) ¾In those frameworks, all languages identify GFs by means of phrase structure positions, i.e. languages are GF-configurational. ¾Some languages have been argued not to be configurational.
¾ Here we focus not so much on the implications that the existence of languages without GF-configurationality could have on framework which assume configurationality to be universal since this has been discussed in detail... but... ¾ We try to illustrate the role of alternative means of GFidentification. ¾ We chose to examine the case of Bulgarian for a couple of reasons...
Bulgarian presents a typologically interesting case for configurationality: ¾Although 80.5% SVO, spoken Bulgarian has a flexible word order. ¾Unlike most Slavic languages and unlike many nonconfiguratioanal languages, Bulgarian has lost its case system. ¾Bulgarian has alternative morpho-syntactic, headmarking means of GF-encoding. ¾In the quite extensive literature, Bulgarian has always been assumed to be configurational. ¾No evidence has been found to confirm the assumed configurationality of Bulgarian;
Outline ¾ First, we adapt and apply configurationality test from King (1995) and Speas (1990) to Bulgarian ¾ We argue that Bulgarian does not provide conclusive evidence for GF-configurationality. ¾ We will also sketch how other means than phrasestructure interact in the encoding of GFs and DFs. ¾ For this, we will focus on the object clitics in the Bulgarian clitic reduplication construction.
Configurationality tests (Speas 1990) Subject-object asymmetries - SUBJ and OBJ have distinct structural positions: - OBJ, lexically governed by V, and SUBJ is not; - SUBJ c-commands OBJ. Tests for a VP constituent which hosts OBJ but not SUBJ.
Lexical government (ECP) In Bulgarian, both SUBJ and OBJ are governed by V - SUBJ and OBJ extract with equal ease in: wh-questions relative clauses topicalization
C-command In Bulgarian, SUBJ does not c-command OBJ: multiple wh-questions - Superiority applies only to the first fronted wh-phrase (e.g. Bošcoviü 1998, 1997, 1993) and even then not always (exceptions involved clitic doubling, work in progress); Can be easier accounted for by NOM-WH > rest binding of pronouns: also explainable by linear precedence; weak crossover: cannot be explained by c-command; only linear precedence;
A VP constituent Bulgarian does not allow: VP-ellipsis, VP-pronominalization, or VP-fronting Coordination: both SUBJ+V and V+OBJ phrases can be coordianted However, SUBJ+V conjunction looks very much like right-node raising.
There is no evidence that Bulgarian is GF-configurational: TEST Extraction Multiple wh-questions RESULT No No Pronominal binding taken alone? Weak crossover Binding and weak crossover No No VP: ellipsis, fronting, pronominalization? VP coordination No/Yes
If not through phrase-structure - how are GFs encoded in Bulgarian? Jaeger & Gerassimova (2002) show that information structure in colloquial spoken Bulgarian is marked by object clitics and prosody. This in turn interacts with the identification of GFs. E.g. the direct object clitic (DOC)... for the current purpose: an optional morphosyntactic agreement marker
Word-order and the DOC E.g. for a sentence with a transitive verb, a subject and an object... Without the DOC: SVO, SOV, VSO and marginally VOS and OVS constituent orders are possible (with different intonations). With the DOC: all six theoretically possible constituent orders can be realized (with different intonations).
Possible word orders without the direct object clitic 692 VXEMHFWYHUEREMHFW Veronica vze parite. Veronica took.3s money.def Veronica took the money. 629 Veronica parite vze. 962 Vze Veronicaparite. ' $??Vze parite Veronica. "296?Parite vze Veronica. 269 *Parite Veronica vze.
Possible word orders with the direct object clitic 6&/92 VXEMHFWFOLWLFYHUEREMHFW Veronica gi vze parite. Veronica CL.3pl took.3s money.de F Veronica took the money. 62&/9 Veronica parite gi vze. 9&/62 Vze gi Veronicaparite. 9&/26 Vze gi parite Veronica. 2&/96 Parite gi vze Veronica. 26&/9 Parite Veronica gi vze.
The role of the DOC The DOC identifies a topical direct object, thus simultaneously encoding a DF (topic) and a GF (object): the DOC can reduplicate the object NP if and only if the object is assigned the information structural function of a TOPIC the DOC agrees in person, number and gender (only in the 3.PS.SG) with the reduplicated object NP. The DOC identifies the direct object even though the object is neither case marked, nor at a specific phrase structure position.
Bulgarian is DF-configurational Bulgarian is discourse-configurational (Kiss 1987 on Hungarian, Nordlinger 1998) DFs in Bulgarian are (at least) partly encoded through structural positions (Rudin 1994, 1997, Rudin et al. 1998, 1999, Embick & Izvorski 1994, 1997). fronted topics (preceding complementizer) fronted, prosodically marked foci (following complementizer) In colloquial spoken Bulgarian, fronted topical object have to be reduplicated by the DOC.
Conclusion No conclusive evidence that Bulgarian is GFconfigurational: this can be taken as an empirical/conceptual problem for most existing accounts of Bulgarian. on the other hand, our data show that configurationality tests may yield unexpected results even for languages which are commonly assumed to be configurational. For spoken Bulgarian, prosody and the DOC (together with subject-verb agreement) provide alternative means for GF- and (before all) DFassignment.
Further research & open questions many.. =) e.g. Is configurationality only a metaphor? Is there any pre-theoretic notion of configurationality What does it mean to assume that not phrase structure but GF-hierarchies or linear order determine e.g. wh-word ordering? In other words: what is phrase structure?