On the syntax of Romanian definite phrases

LQ3HWUD6OHHPDQ +DUU\3HUULGRQ HGV 7KH1RXQ3KUDVHLQ5RPDQFHDQG*HUPDQLF6WUXFWXUH9DULDWLRQDQG&KDQJH $PVWHUGDP-RKQ%HQMDPLQVSSí ON THE SYNTAX OF ROMANIAN DEFINITE PHRASES: CHANGES IN THE PATTERNS OF DEFINITENESS CHECKING ALEXANDRA CORNILESCU1 & ALEXANDRU NICOLAE2 University of Bucharest1 &“Iorgu Iordan − Al. Rosetti” Institute of Linguistics, Bucharest2 Abstract We discuss a type of variation in the pattern of definiteness valuation in Old Romanian (XVIth to XVIIIth century), which has never been noticed before, and examine its significance for the evolution of the DP. In Old Romanian, the suffixed definite article variably occurs either on the first or on a lower [+N] constituent (noun or adjective), so that an indefinite adjective may precede the definite noun. In contrast, in Modern Romanian, it is always the first noun which bears the definite article, while, in case an adjective precedes the noun, the definite article occurs on the adjective. The existence of this lower (definite) article raises several questions (the contexts where it occurs, its significance for the emergence of the enclitic definite article, etc.), to which this paper provides tentative answers. We propose that the existence of a lower definite article combined with a tendency for economy made possible the extension of the use of the article to (pre2nominal) adjectives, gradually leading to stricter conditions in the valuation of definiteness in Modern Romanian (Local Agree). At the same time, the lower article is evidence that the Romanian enclitic definite article originates in a post2posed demonstrative. 1. Aim of the paper. Theoretical assumptions 1.1 The problem It is well known that in Modern Romanian (=MR) the definite article '(u)l is a suffix whose position is fixed: it always occurs on the first N(oun) or A(djective) in the group. This distribution is illustrated in (1): (1) a. c. fata frumoasă girl.the beautiful ‘the beautiful girl’ *fată frumoasa girl beautiful.the b. d. frumoasa fată beautiful.the girl ‘the beautiful girl’ *frumoasă fata beautiful girl.the In Old Romanian (=OR), however, the definite article variably occurs either on the first N or A, or on a lower N, so that another nominal constituent, for instance, an (indefinite) A, may precede the definite N: (2) OR: spre ticăloase cuvintele mele audzul îŃi pleacă… (Cantemir) to vicious words.the my hearing your turn ‘Lend your ear to my vicious words.’ (MR: spre ticăloasele cuvinte ale mele audzul îŃi pleacă…) ALEXANDRA CORNILESCU & ALEXANDRU NICOLAE 2 This type of variation has never been noticed before. It is the aim of this paper to explore it and to examine its significance for the evolution of the Romanian DP. The article which appears on a constituent different from the first N or A will be referred to as “the lower article” for convenience. Before describing the lower definite article, it is appropriate to present our assumptions regarding definiteness checking in MR, so as to later identify the source of the contrast between the two stages in the evolution of the DP. 1.2. Theoretical assumptions 1.2.1 Definiteness as an (un)interpretable feature It is likely that in UG, the D head is uninterpretable [ϕ] and interpretable definite (i.e., [uϕ, idef]), since, in as much as it is interpretable, definiteness is tied to “referentiality”. Thus, definite DPs (proper names, pronouns, definite and demonstrative descriptions) have determined reference (Farkas & von Heusinger 2003, Farkas & de Swart 2007), requiring unique discourse referents. From a syntactic perspective it is the D2layer which secures argumenthood (Stowell 1989, Longobardi, 1994, Giusti 1996, 2005, Borer 2005). From a semantic perspective, in theories like DRT (Kamp & Reyle 1993, and references above), in a D + NP structure, it is the D which introduces the (unique) discourse referent, while the NP supplies a descriptive predicative condition. Following the theoretical suggestions of Pesetsky & Torrego (2007) on the syntax of valuation, definiteness will be considered a nominal property, uninterpretable on the noun ([udef]) and interpretable (though unvalued) on the determiner ([idef]). Yet, definiteness may be valued on certain types of nouns, i.e., certain categories of nouns may be marked as [u+def] from the lexicon. In UG, proper names are inherently [+definite] and value the [idef] feature of D, as proposed in Longobardi (1994). Similarly, we propose that in languages where nouns morphologically vary for definiteness, like Romanian, nominal morphology may supply the value of the definiteness feature in D. Concluding, definiteness in D is interpretable and unvalued [idef], and it will be valued either by external merge of a lexical determiner or by internal merge of an NP/DP which is morphologically definite, such as a noun suffixed by the definite article. Assuming that feature valuation is consequent upon external merge of the article, definiteness valuation for a language that has free2standing definite determiners like English might look like the following: (3) DP D [uϕ] [i+def] the NP [iϕ] rose The D head agrees with the N head valuing its ϕ2features. At the same time, the definite article values the [idef] feature of the D head. In agreement with other analysts, we assume that in MR, the enclitic article '(u)l is a suffix (cf. Ortmann & Popescu 2000 among many.), subcategorized for an N—/A— complement, with which it merges in the lexicon: ON THE SYNTAX OF ROMANIAN DEFINITE PHRASES (4) a. trandafirul (frumos) rose.the (beautiful) “the beautiful rose” b. 3 frumosul trandafir beautiful.the rose “the beautiful rose” Let us detail the mechanism of definiteness valuation. When the N is suffixed with the definite article, the result is a definite noun, therefore an NP which is valued for definiteness, marked [u+def, iϕ]. These features of the N are used to value the corresponding interpretable (but unvalued) features of the D head, as shown below: (5) DP D [uϕ] [idef] NP [iϕ] [u+def] trandafirul Agree As mentioned, the definite article may also be suffixed to an A2head. Importantly, MR As may be suffixed with the definite article only when the AP is attributive and pre'nominal; in such cases, the A merges as a specifier of the NP, so that the adjectival head is in a configuration of local Agree with the NP: (6) frumosul trandafir beautiful.the rose “the beautiful rose” When the A is predicative (e.g., a post2copular predicative (7a) or a post2nominal adjunct (7b)), the definite article is impossible: (7) a. *Trandafirul este frumosul. rose.the is beautiful.the b. *trandafir(ul) frumosul rose(.the) beautiful 1.2.2 Locality of Move and Agree in Modern Romanian DPs Definiteness is therefore an agreement feature for A. We assume that MR As, by virtue of being ϕ2complete, may also bear an [udef] feature which is never valued by As themselves, but may be valued by a definite N. The A probes the nominal that it c2commands and will agree with the N in definiteness and ϕ2features, so that the A, which enters the derivation [uϕ, udef], may end up being [uϕ, u+def], its features being thus identical with the N’s. When this happens, phonology always realizes [u+def] on the highest copy below D, i.e., the highest N or A below D, which will bear the definite article at PF. This highest copy is the one that values the [idef] feature of the D head. The process of definiteness feature transmission assumes the form of a series of Agree relations ((8b), (8c)); finally, the definite A immediately below D values the [idef] feature of D (8c). (8) a. frumosul trandafir beautiful.the rose “the beautiful rose” ALEXANDRA CORNILESCU & ALEXANDRU NICOLAE b. 4 FP AP F’ F NP A N [iϕ] [u+def] trandafir [uϕ] [u+def] frumosul Agree c. DP D FP AP [uϕ] [i+def] F’ A F [uϕ] [u+def] NP N [iϕ] [u+def] trandafir frumosul Agree The description given above shows that in MR Agree is strictly local, as in (9): (9) Definiteness valuation in Modern Romanian − Local Agree (LA)): The [+def] GoalP which values [idef] in D must be a [+N] phrase immediately below D. A definite AP/NP must end up immediately below the D, in a position where the [u+def] feature of the definite N/A is accessible to D head for strictly Local Agree. This is why (1c) and (1d) are impossible. The article is too remote to function as a Goal for the probe in D. There is an opaque intervener (the NP in (1c), the AP in (1d)) between the Probe and the Goal. Pairs like (1a) and (1b) (=(10)) represent different derivations ((10a) has the derivation in (8c) above; (10b) has the derivation in (11) – irrelevant details aside), both observing the same generalization in (9): (10) a. frumosul trandafir beautiful.the rose ‘the beautiful rose’ (11) b. DP D [uϕ] [i+def] NP NP N [iϕ] [u+def] trandafirul Agree AP A [uϕ] frumos trandafirul frumos rose.the beautiful ‘the beautiful rose’ ON THE SYNTAX OF ROMANIAN DEFINITE PHRASES 5 We conclude that in MR, it is the first [+N] constituent (N or A) of the DP which values the feature in D, by Local Agree. Romanian and French contrast with English in the following modifier + proper name structure: (12) a. c. le vieux Paris the old Paris “old Paris” old Paris b. vechiul Paris old.the Paris “old Paris” Unlike the two Romance languages, English allows Long Distance Agree (LDA), in the sense that the N which values the feature in D may be separated from D by an intervening adjective. French and Romanian disallow LDA, but use alternative strategies to value definiteness: a free standing definite article (French) or a definite article suffixed on an A immediately below D (Romanian). Locality of Move For the analysis that follows, it is also important to mention one more property of the MR DP related to the post2nominal Dem(onstrative) and, more generally, to the application of DP2internal Move. We claim that Move, like Agree, is local; it cannot cross a higher specifier. As to demonstratives, like Spanish, MR disposes of both pre2nominal (13a) and post2nominal Dems (13b). The former are similar to their Romance counterparts and need no discussion. The latter have special locality constraints, namely the post2nominal Dem can only be preceded by a definite N, strictly adjacent to it (13b) (see the ungrammaticality of (13c) and (13d)). (13) a. acest (foarte) frumos portret this (very) beautiful portrait “this (very) beautiful portrait” c. *portretul (foarte) frumos acesta portrait (very) beautiful this b. portretul acesta (foarte) frumos portrait this (very) beautiful “this (very) beautiful portrait” d. *(foarte) frumosul acesta portret (very) beautiful this portrait Strict adjacency of the post2nominal Dem to the definite head has standardly been interpreted as an instance of definite N02Movement to D across a phrasal Dem in the specifier below D, as shown in the intermediate configuration (14) (cf. Cornilescu 1992, Vulchanova & Vulchanova1998 among many). (14) DP D’ D [idef] [u2ϕ] [+EPP] FP DemP Dem F [u+def] [uϕ] F’ NumP Num NP N [u+def] [iϕ] portretul acesta ALEXANDRA CORNILESCU & ALEXANDRU NICOLAE 6 The examples in (13) show a sharp contrast between APs and NPs in the demonstrative construction. In contrast to definite Ns, definite As cannot move across the DemP, as known since Grosu (1988). In MR, As always move as phrases (14d) and, if Move is local, APs cannot move past the demonstrative specifier. This raises the issue of why N2Raising is possible, while AP2Raising is not. Notice that the DemP and the definite NP/AP share their [ϕ] and [+def]2features, which makes them equally good goals for the probe in D. Consequently, it is the DemP which is closer to D that should value definiteness. This is what blocks the raising of the AP, as expected. Once again, the definite NP and the DemP are equally good goals for probe in D, but the N2head is lighter and as such is preferred as a mover. The preference for N02Movement thus follows from a principle of economy which requires piping only as much material as is necessary for convergence (Pied Pipe Less Weight − Stateva 2002, or Attract/Move Smallest − Akiyama 2004). The N successive2cyclically moves crossing the DemP and ending up in D. Conclusion What has been said so far proves that the syntax of the MR DP shows locality constraints on both Agree and Move. 2. Long Distance Agree in Old Romanian 2.1 The Lower Definite Article Against the background sketched in Section 1., we may now turn to the variation in the pattern of definiteness valuation in OR (XVIth to XVIIIth century), which we mentioned above and which has gone unnoticed so far; no mention of it is made of it in important histories of Romanian (GheŃie 1975, Rosetti 1968, Densusianu 1961, Dimitrescu, 1978, BrancuT 2004, Niculescu 1990). In OR, the definite article variably occurs either on the first [+N] constituent or on the head N in a lower position; another nominal constituent, for instance, an A, will precede the definite N in such cases: (15) ca mare scâr9netul roatelor like great grinding.the of the wheels ‘like the strong grinding of the wheels’ (Cantemir) Examples of this type show that locality conditions are not so strict in OR, allowing a different pattern of definiteness checking: (16) Definiteness valuation in Old Romanian − Long Distance Agree (LDA): The goal that values the probe in D is a c2commanded nominal phrase (NP, AP) which need not be the first (nominal) phrase c2commanded by D. This amounts to saying that the [+def] feature is realized either on the first or on a lower nominal constituent of the DP. The definite article which shows up on a N which is not the first constituent of the DP is what we have called the “lower definite article”. ON THE SYNTAX OF ROMANIAN DEFINITE PHRASES 7 2.2 The extension and range of the Lower Definite Article The lower definite article is present from the oldest Romanian texts of the XVIth century up until the first half of the XVIIIth century. This phenomenon thus belongs to OR, as opposed to MR, assumed to start at the end of the XVIIIth century (cf. GheŃie 1975): (17) a. ...au venit egumenul de BistriŃă cu cinstită cartea mării tale (XVIth – DÎR) has come abbot.the of BistriŃa with honoured letter.the highness.theGen your “...the Abbot of BistriŃa came with your highness’ honoured letter” b. au aflat cap 9i începătura mo9ilor [...] ca să nu se înece (they) had found head and beginning ancestors.the.Gen so that not be drowned trecuŃi (XVIIth, 1641 – Ureche ) a toate Ńările anii of all countries.the years.the passed “They found the origin and the beginning of their ancestors so that the passed years of all countries may not be drowned into oblivion” Măriei Tale, Radu logofăt... (XVIIth, 1688 – Biblia) c. plecat robul humble servant.the highness.the.Gen Your, Radu Chancellor “your highness’s humble servant, Radu Chancellor” (XVIIth, 1671 – Costin) d. Umblăm după a lumii în"elătoare faŃa Go we after ALgenitival article world’s deceitful face.the “We are after the world’s deceitful face” e. A9a, fără veste el în vrăjma"i colŃii crocodilului aflându'se (XVIIIth –Cantemir) thus suddenly he in inimical teeth.the crocodile.the.Gen being “Thus, suddenly, he was in the crocodile’s inimical teeth” As to the range of the constituents that may precede the lower definite article suffixed on the N, they may be of two types: As and pre2nominal genitives (Gens). (18) (19) (A) an (indefinite) adjective A9a, fără veste el în vrăjma"i colŃii crocodilului aflându'se (Cantemir) thus suddenly he in inimical teeth.the crocodile.the.Gen being “Thus, suddenly, he was in the crocodile’s inimical teeth” (B) Gen Phrase (Gen DP) (Ureche) ...ca să nu se înece a toate Ńările anii trecuŃi so that not be drowned of all countries.the years.the passed “…so that the passed years of all countries may not be drowned into oblivion” One should know that in MR a pre2nominal Gen is followed by an indefinite N. Just as in English, a DP containing a pre2nominal Gen is interpreted as [+def], and it is the pre2nominal Gen DP which checks the [idef] feature of the DP. The pre2nominal Gen DP in MR functions as a definite Determiner Gen (Huddlestone & Pullum 2002). It also occurs in the first position of the DP, presumably in [Spec, DP]. (20) a. al regelui fiu ALgenitival article king.the.Gen son “the king’s son” b. fiul regelui son.the king.the.Gen “the king’s son” ALEXANDRA CORNILESCU & ALEXANDRU NICOLAE 8 Examples like (19) contrast with MR (20a), and are no longer found in MR. The lower definite article raises several questions: 1. What is the interpretation of this phenomenon in the framework sketched above? 2. What made the lower article likely or possible in OR? 3. What are the contexts that favored the occurrence of the lower article? 4. What are the causes that led to the elimination of this pattern? 5. Are there other OR DP structures which relate to the existence of LDA? In the following sections of the paper we supply tentative answers to these questions. 2.3. Interpreting the facts of Old Romanian in the framework sketched in Section 1 Two morpho2syntactic properties of Romanian N/A have combined to produce the strict locality conditions on definiteness valuation in MR. The first is the suffixal nature of the article which allows Ns to be valued for definiteness, even if their definiteness feature is uninterpretable (i.e., [u+def]). The second significant property is that, at some point in the evolution of Romanian, As must have acquired the possibility of optionally incorporating an uninterpretable unvalued definiteness feature [udef]. This feature was valued through Agree with the N, as shown above. Since Agree was long distance, and the nominal valued for definiteness did not need to be the first NP/AP below D, we expect the following alternations in OR, all of which are attested: (i) The definite article shows on the first NP or AP of the DP (the MR pattern, available in all attested stages of Romanian) (21) au purces fără numai din vechea "i rânceda (it) happened only out.of old.the and rancid.the “It all happened out the old and rancid envy” pizmăluire (Cantemir) envy (ii) The definite article shows on an NP which is not the first phrase of the DP (this is the lower article) (22) spre ticăloase cuvintele mele audzul îŃi pleacă… (Cantemir) to vicious words.the my hearing your turn ‘Lend your ear to my vicious words’ (iii) Sporadically, the definite article could also be present on more than one constituent, i.e., the DP shows multiple definites (cf. Croitor 2008). This is consistent with the view that definiteness had become a concord features in OR, and Romanian morphology allows it to be uninterpretable but valued and realized on both Ns and As. Multiple definites are present both in the order A+ N (examples (23), from Croitor 2008), and in the order N+ A (examples (24), likewise from Croitor 2008); apparently, multiple definites were lost (in these patterns) at the end of the XVIIIth century: (23) (A) Multiple definites: A+N a. Ce i'au tăiat atuncea curând puternica mâna lui Dumnedzeu zilele (Costin) that to2him have cut then soon mighty.the hand.the of God days.the “God’s almighty hand took his/her days” ON THE SYNTAX OF ROMANIAN DEFINITE PHRASES 9 b. …moa9tele a sfintei prepodobnei Paraschevei, în (Costin) …relics.the of saint.the beautifully2adorned.the Parascheva, in “…the relics of the holy, beautifully2adorned (Saint) Parascheva” (24) (B) Multiple definites: N+A "i frumoasei în Spania zidul cetăŃii marei wall.the city.the.Gen big.the.Gen and beautiful.the.GEN in Spain “the wall of the big and beautiful city in Spain” (Cantacuzino) It is reasonable to assume that it was precisely the possibility of valuing definiteness on pre'nominal As that led to stricter locality conditions on Agree, that is, to the MR requirement that the constituent that values the [idef] feature of D should be the first AP/NP below D. Gradually, as a result of a general tendency towards economy, the constituent that values [idef] in D (i.e, which has an LF effect) got to be the only one which phonologically realizes definiteness. This was either a definite pre2nominal A or a definite N. Multiple occurrences of the suffixal article are now ruled out. Moreover, only pre'nominal As are ever suffixed by the definite article, since only pre2nominal As can be closer to D than the N. Definiteness thus turns into an exclusive property of attributive As as opposed to predicative ones (see above). This means that As that merge as attributes (specifiers or pre2nominal adjuncts) are obligatorily specified as [udef]. This is what guarantees that if the head N is [u+def], and there is a pre2nominal A in the DP, definiteness will be realized on the pre2nominal adjective immediately below D. 3. What the Lower Article suggests about the emergence of the enclitic article 3.1 An open question The account proposed above relies on the contrast between languages which value definiteness in D by merging a determiner and languages which value D by means of a lexical category morphologically marked as valued for definiteness. From a Romance diachronic perspective, one of the still hot puzzles is the manner in which the same Latin demonstrative ille (cf. Iordan & Manoliu 1965) led to proclitic free2standing articles in other Romance languages, but to an enclitic article in Romanian. We claim that the lower article tilts the balance for the hypothesis that the enclitic article developed out of the post2nominal Latin Dem ille rather than out of a pre2nominal demonstrative, as in Giusti’s analysis of the history of the definite article in Romance. Both proposals have been advanced for Romanian as well (see Renzi 1993 for the pre2nominal demonstrative hypothesis, and Coteanu 1956 and Graur 1967 for the post2nominal demonstrative hypothesis). The lower article matters in this on2going debate, since it is the so far only attested construction that does not show the Romanian definite article on the head of the first NP/AP of the DP. The standard “high article” is consistent with analyses where the article merges in D and is lowered on the first [+N] constituent (as recently proposed by Dobrovie & Giurgea, 2006), or where the article merges in D and there is movement into ALEXANDRA CORNILESCU & ALEXANDRU NICOLAE 10 the D2area (N2to2D, AP2to2Spec, DP) as assumed in older studies (Dobrovie2Sorin 1987, Grosu 1988). The lower article should be construed as evidence that the article merges low, and “moves” higher through Agree reaching the position below D. The aim of this section is two2fold: we propose a possible path from the Latin post2nominal demonstrative to the enclitic definite article; secondly we argue that, if correct, the diachronic proposal is consistent with an analysis of the enclitic article as a suffix rather than a (second position) clitic. Before detailing the change from post2nominal Dem to enclitic definite article, we spell out our (minimal) hypotheses regarding the architecture of the DP. Following a number of researchers (Borer 2005, Giusti 1993, Julien 2005, Roehrs 2006), we assume that articles, or rather, determiners, merge lower, say in an Art(icle)P (valuing the [ϕ] and [idef] features of the Art head), and then move to D or [Spec, DP] to value the features there (deixis, specificity, definiteness, [uϕ]), if we assume, as mentioned above (cf. also Giusti 1993, Longobardi 2001) that an argument DP is interpretable only if its D projection is visible. The space between DP and ArtP may (but need not) host periphery or quantificational projections, as in the proposal put forth by Roehrs (2006): (25) DP > CardP > ArtP > NumP > NP It follows that the Dems merge in Spec, ArtP, under the assumption that they are phrasal. We also accept that pre2nominal As in Romance merge as specifiers of functional projections (Bernstein, 1991, 1993), while post2nominal As merge as adjuncts (Giurgea 2008). As shown, MR relies on LA and Local Movement. OR appears to have employed LDA (checking of a feature across an intervening specifier), and Long Distance Movement (movement over a relevant specifier). In agreement with proposals for Scandinavian (see especially Roehrs 2006: 49264) and Romance (Renzi 1993, 1997, Lyons 1999, Giusti 1998), we might envisage the following scenario for the rise of the enclitic article. The enclitic article emerged through the reanalysis of the Dem ille in the context of the post'nominal Dem construction; reanalysis amounted to a change in its c2selection properties. This hypothesis is supported by the existence of post2nominal Dems in all written phases of Romanian (26): (26) a. Au trimis Pa9a pre tălmaciul acela have sent Pasha PE translator.the that “Pasha sent that translator” (Costin) b. 9i au căzut în războiul acela and have fallen (they) in war.the that “And they fell dead in that war” (Ureche) The post2nominal position of As, possible in Latin, was reinforced by the contact with the local Dacian idioms, which strongly preferred to post2pose the A, including the (adjectival) Dem (BrâncuT 2004, Graur 1967). In all attested stages of Romanian, the pre2nominal or post2nominal position of the Dem depended on its textual, prosodic role (deictic or anaphoric). Consequently, it may be believed that either the NP or the DemP alternatively moved to [Spec, DP] to check the unvalued features there, thus deriving the alternative orders ((27a) and (27b)). ON THE SYNTAX OF ROMANIAN DEFINITE PHRASES (27) a. DP NP D’ D [+def] ArtP DemP Art’ Art homo [−foc] b. ille [+foc] NumP Num’ tNP DP DemP D’ D ArtP Art’ t DemP Art ille 11 NumP NP homo Num’ As known, in two2member structures like omul acesta (‘man.the this’) the post2posed Dem is normally focused (Bernstein 2001 a.o.), and may have a [locative2deictic] feature, while the Romanian pre2nominal Dem (acest om ‘this man’) is mostly anaphoric (Manoliu2Manea 1993). It is, then, more plausible to assume that the re2analysis of the Dem as an article occurred in a three'term construction, which would shift the nuclear stress on the last (most deeply embedded) third term (Cinque, 1993). Graur (1967) thus suggested as a basis for re2analysis the three2term construction: homo ille bonus, N + Dem + Adj. Since the Dem is not in focus, and thus probably not stressed, it is likely that it was “weakened” and re7analyzed as a head moving to D0 rather than to [Spec, DP], following the general evolution of Romance or Germanic (Roehrs 2006 for Scandinavian). It is true that, as remarked by one of the reviewers, since in the earliest Romanian texts the article is already fully grammaticalized, there is no decisive piece of evidence that re2analysis occurred in the three term structure. The demonstrative might perhaps have been distressed as a result of its anaphoric function in the simpler two2term construction as well. The change from a phrase to a head moving to the D2head position would have been the same. The often made suggestion that re2analysis is based on structures including post2nominal modifiers probably takes into account the fact that the definite article is still required to license a post2nominal modifier/argument in prepositional constructions, where the nominal head is otherwise determinerless (Isac 2006): (28) a. Cartea este pe masă. book.the is on table “The book is on the table” b. Cartea este pe masa rotundă. book.the is on table.the round “The book is on the round table” ALEXANDRA CORNILESCU & ALEXANDRU NICOLAE 12 Head movement of Dem to D frees [Spec, DP] for NP2movement; suffixation takes place in this Spec2Head configuration. Re2analysis of the Dem as a suffix represents a change of its c2selection feature, which becomes [+N222], satisfied by combining it with a nominal head (an N0, A0 constituent). Head2adjunction of the Dem2article to the N causes the former to undergo phonological reduction, dropping its first syllable ILLE '(U)L. Significantly, the pre2nominal definite article of French, which has developed out of a pre2nominal demonstrative (Iordan&Manoliu 1965, Giusti 1993, 1998), has also further developed from an independent head into a prefix in some of the creoles based on French (Mauritian Creole), being reanalyzed as part of the noun stem (Lyons 1999: 331; examples from Lyons 1999): (29) a. b. le lit the bed ‘the bed’ lili la the.bed there ‘that bed’ (Standard French) (Mauritian Creole) The change from Dem to article also meant a loss of semantic features (bleaching), from the richer matrix of the Dem [(locative)2deictic, definite, anaphoric, 3rd/6th person, adjective, /pronoun] to the more reduced feature matrix of the article [definite, anaphoric, adjective] (cf. Giusti 1998, Lyons 1999). (30) DP NP N D’ D ArtP Dem+D DemP homo (il)le om+(u)l t Dem Art’ Art NumP NP Num’ tNP AP bonus Num.... …..bun The lower article preceded by an A is a precious missing link in the chain leading from a post2nominal Dem to an article placed on the first nominal constituent of the DP. Thus Renzi (1993: 308), expressing skepticism about the post2nominal Dem hypothesis, was wondering: “Why starting from omul bătrân [man.the old ‘the old man’], we have bătrânul om [old.the man ‘the old man’], and not bătrân omul [old man.the]?”, in other words why is there no evidence of an article on a lower term than the first ? The examples that we have surveyed are exactly of the type expected by Renzi (1993) to occur under the hypothesis of deriving the article from the post2nominal Dem. ON THE SYNTAX OF ROMANIAN DEFINITE PHRASES 13 3.2 Consequences for the analysis of the article: the Romanian definite article is a suffix The historical path suggested above is consistent with the claim that the combination of the article with the N or A is the outcome of a morphological rule, rather than the effect of syntactic movement; in particular, it is not the output of N2to2D or AP2to2Spec, D, as previously believed, since the important work of Dobrovie2Sorin (1987) and Grosu (1988). Strong evidence shows that the definite article is a suffix (Ortmann & Popescu 2000, Dobrovie2Sorin & Giurgea 2006, Tasmowsky 2009), rather than a second position clitic / a Wackernagel clitic (Renzi 1993). The morphologic status of the article as a suffix rather than a clitic follows from the following facts: (i) the article is not always in second position, since, inside the AP, degree words (and prepositions) may precede the A to which the article attaches, as in atât de / foarte lungi+le drumuri (‘such of / very long.the roads’); (ii) the article attaches to both conjuncts in a coordination, as in frumos+ul 9i mare*(+le) ora9 (‘beautiful+the Ti big*(+the) city’), an unexpected repetition for a clitic (Zwicky & Pullum 1983); (iii) the article shows allomorphy, changing its form function of the last phoneme and the inflectional class of the stem; (iv) the article occurs in a constant position, i.e., on the first N or A in the DP. In contrast, Romanian clitics are inconsistent with respect to their position (since they may occur both before and after their syntactic host: l'am luat/ am luat'o ‘himCL2haveAUX, 1st person taken’ / ‘haveAUX, 1st person taken2herCL’), as well as to the type of constituents they cliticize on (verbs, auxiliaries, Ns, complementizers). 4. Contexts of occurrence of the Lower Definite Article A relevant question, already formulated above, is what contexts require or allow the use of a lower article. To answer this question, a body of texts ranging from the earliest Romanian writings of the XVIth century to the first half of the XVIIIth century was examined. The texts show variation between the ‘high’ article, used in most cases, and the ‘lower’ article. The lower article predominantly occurs when the definite head N is followed by another constituent (especially by a Gen(itive)). It is for modified and complemented (by a Gen) DPs that we have checked the relative frequency of the high vs. lower article (see (31) below). (31) Text Miron Costin LetopiseŃul Cantacuzinesc R. Popescu R. Greceanu Constantin Cantacuzino Cantemir TOTAL ADEF + N + Gen DP (high article) 0 3 0 2 0 1 6 High article: 6 (12,5%) A + NDEF + Gen DP (lower article) 15 1 1 3 5 17 42 Lower article: 47 (87,5%) The examination of the data shows that a lower definite article on N appears overwhelmingly (87,5%) in contexts where the head N is followed by a Gen DP complement. Other post2nominal modifiers may also sometimes trigger the presence of ALEXANDRA CORNILESCU & ALEXANDRU NICOLAE 14 the lower article as in (34). If there is no complement or modifier, the article emerges on the pre2nominal A, as in MR. (32) (33) (34) (A) A[+def] + N[7def] (no complement/modifier) au purces fără numai din vechea "i rânceda pizmăluire (Cantemir) (it) happened only out.of old.the and rancid.the envy “It all happened out the old and rancid envy” (B) A[ 7 def] + N [def] + GenDP roatelor (Cantemir) a. …ca mare scîr"netul like great grinding.the of the wheels “…like the strong grinding of the wheels…” b. …Corbul de uscate vinele goalelor ciolane clonŃul 9i'ar ciocni (id.) the Raven against dried2up veins.the empty.theGen bones bill hid would knock “The Raven might knock his bill against the dried2up veins of his bones.” c. ...de dulce otrava Hulpii tare se ameŃiră (Cantemir) …with sweet poison.the of the Vixen much (they) got drunk “…they got quite drunk from the sweet poison of the Vixen” (C) A[ 7 def] + N [def] + Modifier (PP or AP modifier) despre vrăjitorii vremii trasă (Cantemir) a. însă nu puŃine asuprele but not few injustices.the from the magicians of the times (he) endured. “...but he endured many injustices from the magicians of the times…” b. Ne"tiutor gândul omenesc […] la ce merge..? (Cantemir) Ignorant thought.the human what is heading for? “What is the ignorant thought of man aiming at?” This distribution signals a tight relation between the inflectional Gen (and other modifiers) and the lower article, a fact which should be accounted for. In MR, i.e., after 1780, the lower article disappears. There are, however, two types of motivated exceptions. The lower article is still part of religious and other obsolete texts (35), and in such case it is again mostly followed by the Gen. Secondly, there is a small group of quantificational or evaluative As that may function as definite quantifiers, and may or must be followed by definite Ns (see also GALR 2005, Barbu 2004): întreg (‘whole’), singur (‘unique’) ditamai (‘big’), gogeamite (‘big’), as in (36). No Gen modifier is required. (35) (36) a. Miluie9te'mă, Doamne, după mare mila Ta! Have2mercy2on2me God, according to great mercy.the Your “God, have mercy on me, according to your great mercy.” b. ditamai prostul / *ditamaiul prost întreg ora"ul / întregul ora" whole city.the whole.the city big fool.the We may conclude that one significant change in the syntax of the Romanian DP relates to the locality conditions of definiteness valuation. The definite feature strengthens requiring to be valued by a strictly local nominal constituent (N or A). Long Distance Agree is ruled out. ON THE SYNTAX OF ROMANIAN DEFINITE PHRASES 15 5. The (Lower) Article and the reorganization of the Genitive system 5.1 The inflectional and the prepositional Genitive Romanian differs from other Romance languages in that it disposes of an inflectional Gen, while in other Romance languages the Gen is prepositional, marked by de ‘of’ (Grosu 1988 a.o.). The creation of a nominal inflectional Gen in Romanian was often viewed as an ‘effect’ of the suffixal definite article. Thus, an influential traditional opinion was that “in Romanian, the maintenance of oblique cases is the first and most important cause of the post2position of the definite article” (Coteanu 1956: 67). As also underlined by one of the reviewers, there is no demonstrable correlation between enclisis and the development of a inflectional Gen, or the other way round. It is true to say, however, that the development of the article system favored the reorganization of the Gen system: the Gen case has developed parallel, inflectional and prepositional forms (see (37)), and it is the suffixal article which varies for Case in Romanian. Thus not only did Romanian develop an inflectional Gen, as is well known, but the prepositional Gen, based on the same preposition DE as in all Romance, did not disappear, either. Rather it became very limited and specialized (see Cornilescu 2004 for details). In OR, the inflectional and the DE Gen are in free distribution (at least in post2nominal position where both occur) as shown by Pană Dindelegan (2008). Romanian has developed a morphological distinction between “anchoring Gens”, always DPs, and “non2anchoring (Prepositional) Gens”, always syntactic NPs (in the sense of Koptjevskaya2Tamm 2005), thus verifying the typological generalization that only languages that have articles may develop specialized forms for anchoring vs. non2anchoring Gen. The two forms show a very different cluster of morpho2syntactic and semantic properties, summed up below: (37) (38) a. citirea reading.the b. citirea reading.the Anchoring Gens cât mai des a autorilor clasici more frequently of classical authors frecventă de romane poliŃiste frequent of crime fiction Non'anchoring Gens (limited to nominalizations, see below) a. inflectional a. prepositional b. DP b. NP c. referential, < e>2type denotation c. <e, t> denotation In sum, Romanian has developed a reliable syntax2semantics correlation in the domain of the Gen, by developing an inflectional Gen system alongside of the prepositional DE one. 5.2. Significance of the statistical correlation between the Lower Definite Article and the inflectional Genitive As noticed in table (31), there is a strong statistical correlation between the lower article and the inflectional Gen. This correlation may be interpreted from a double perspective: ALEXANDRA CORNILESCU & ALEXANDRU NICOLAE 16 (i) One may adopt a functional2pragmatic perspective, considering as significant the referential interpretation of the inflectional Gen, as opposed to the property, generic interpretation of the prepositional Gen. One might say that the lower article always accompanied that Gen form which served as a contextual anchor for the head. The article, initially a post'posed Dem, had a similar anchoring role, probably being interpreted as [locative2deictic] (Lyons 1999, Brugé 2000, Coteanu 1956, Manoliu2 Manea 1993). The functional role of the post2posed article may have been that of a weak Dem, i.e., a form intermediate between a Dem and a definite article (Lyons 1999, Giusti 1998). It is the correlation between the anchoring role of the Gen and the presence of the post2posed Dem which is still visible in the lower N [+def] examples noticed above. (ii) The correlation between the lower article and the post2posed inflectional Gen can also be viewed from a more narrowly distributional perspective, more likely to be the correct one It is known that the inflectional Gen in Romanian is realized either by a bare inflected DP (= the bare Genitive (39b)), or as an inflected DP preceded by the genitival article AL (39a) (for a description of the Gen article see Cornilescu 1995, 2005). The two forms are in complementary distribution. Roughly, the AL Gen occurs whenever the head N is indefinite (39a), while the bare Gen occurs when the head N is definite and the Gen is adjacent to the head N (39b): (39) a. b. doi prieteni ai copilului two friends ALgenitival article child.theGen “two friends of the child” prietenul copilului friend.the child.theGen “the child’s friend” It is the syntax of definite DPs containing Gens that is of interest. With definite heads, the AL Gen occurs in two situations: (i) when there is an intervening modifier between the definite head and the Gen (40a); (ii) when the Gen is adjacent to the head, but the article is on a pre2nominal A (40b). (40) a. b. c. prietenul bun al copilului friend.the child.theGen “the child’s good friend” bunul prieten al copilului good.the friend ALgenitival article child.theGen “the child’s good friend” *bunul prieten copilului good.the friend child.theGen Generalizing, the bare Gen occurs only when it immediately follows a definite N, while the AL Gen occurs elsewhere. The bare Gen is the preferred form since it is more economical (economy of representation). The preference for the bare Gen clearly must have been felt in OR as well. This is what explains the use of the lower definite article with inflectional Gen. If the article is placed on the N, instead of being placed on the ON THE SYNTAX OF ROMANIAN DEFINITE PHRASES 17 higher A, the Gen is adjacent to the definite article and it is possible to employ the more economical bare Gen (41a) (= 17a)), replacing the AL Gen (41b): (41) a. OR: b. MR: cu cinstită cartea mării tale (Bare Gen) with honoured letter.the highness.theGen your (AL genitive) cu cinstita carte a mării tale with honoured.the letter ALgenitival article highness.theGen your Thus both formal economy considerations and functional semantic considerations may be invoked to account for the preference for the lower article in the context of an inflectional Gen. 6. Other OR DP patterns where Agree and Move operate Long Distance In this section we show that the possibility of valuing the [idef] feature in D across a specifier, that is, LDA, was very general in OR and it combined with long distance move, therefore with movement across a specifier, a fact which is no longer allowed in MR. These hypotheses allow us to understand a number of other structures possible in OR, but systematically excluded in MR. So far we have only examined cases where LDA is expressed by a lower article. There are, however, other constituents that may value D2definiteness, such as Dems or inflectional Gen DPs. In OR, these types of constituents could, like the definite article, value D2definiteness either by LA, or by LDA, i.e., at a distance from the D position. Variation in the pattern of definiteness valuation thus also had consequences for DPs that did not include the definite article, but which show Dems and Gen DPs in constructions no longer available in MR. In this section, we review some of these constructions which involve LDA or LD move. 6.1. As shown in Section 1., the post2nominal Dem cannot be preceded in MR by As. In contrast, in OR, both indefinite and definite As could precede the Dem. Consider indefinite As first: (42) a. OR: b. MR: cumplite aceste vremi de acmu terrible these times of now “these terrible times of now” aceste cumplite vremi de acum these terrible times of now The A in (42) is surely in an emphatic periphery position. As proposed by Laenzlinger (2005), periphery As merge or move to the left periphery of the DP defined by him as the space between a lower DAgreement and a higher DDeixis. This proposal is rather similar to Roehrs’s (2006) in (25) above, who also proposes that Agreement features are checked in ArtP, while ‘referential’/ deictic features are checked in the higher D position. Under these assumptions, an example like (42) would have the structure in (43): the Dem merges in [Spec, ArtP], the A is above it in a periphery projection, and the Dem checks its deictic feature across the A by LDA. ALEXANDRA CORNILESCU & ALEXANDRU NICOLAE (43) 18 DP D [idef] EmphP AP Emph’ Emph cumplite ArtP DemP Art’ aceste [u+def] Art NumP NP…. vremi The absence of this pattern in MR is the result of the disappearance of LDA. 6.2 In OR, it is also possible for a complex definite NP to precede the Dem, contrary to MR, where the N0 alone crosses the Dem (see section 1.). (44) a. b. pă ticălosul pământŭ acesta să vie on wretched.the earth this come.Subj “…that he should come on this wretched earth..” inima ta aceasta heart.the your this “this heart of yours” (Greceanu) (Greceanu) The derivation of these examples involves movement of a phrase across a specifier, i.e., LDM. A relevant intermediate structure is the one in (45). Assuming that there is NP movement (at least) to NumP in Romance (cf. Cinque 2004 among many authors) the definite NP is in [Spec, NumP] functioning as a Goal for the unvalued [ϕ] and [def] features of the Art head. The Dem merges in [Spec, ArtP], valuing the features of the higher D, through LDA. The definite NP moves to [Spec, DP] presumably to avoid the focus interpretation. It is apparent that in moving to [Spec, DP], the definite NP crosses a phrasal constituent in [Spec, ArtP], this being an instance of LDM. (45) DP D [i+def] EmphP Emph ArtP DemP acesta Art’ Art [+phi] [+def] NumP … FP AP ticălos+ul F’ F NP pământŭ ON THE SYNTAX OF ROMANIAN DEFINITE PHRASES 19 Expectedly, the structure in (45) is impossible in MR, where Move is strictly local. Notice also the examples below in (46), where a definite A alone has moved across the Dem, after checking definiteness against the ArtP head, as in (45) above. (46) a. OR: nenorocitele acestea vremi unfortunate.the these times ‘these unfortunate times’ (MR: vremile acestea nenorocite) (Greceanu) 6.3 A second example of LDA in DPs that do not contain the definite article is offered by pre2nominal Gen DPs. With pre2nominal Gen DPs, as with Dems, definiteness is checked either at a distance or in a local configuration. First, there are DPs where the Gen is pre2 nominal but is not involved in valuing definiteness. A definite determiner (the definite article, a Dem) or an indefinite one values the feature in D. In such cases, the pre2nominal Gen has nothing to do with the checking of definiteness (the valuer of the [idef] feature in D is the definite article (47a), the Dem (47b) or an indefinite determiner (48)) (47) a. ca să nu se înece a toate Ńările anii trecuŃi (Ureche ) so that not be drowned of all countries.the years.the passed “[…] so that the passed years of all countries may not be drowned into oblivion” b. aceste ale Ciacalului cuvinte (Cantemir) these AL Jackel.the.Gen words “these words of the Jackel” (48) acele jigănii, carele [...] într'altă a trupului parte arme [...] poartă (Cantemir) those beasts, which […] in other AL body’the.Gen part arms [] carry “those beasts which carry [] arms in some other part of their body” Interestingly however, a Gen DP may incorporate a [+def] feature and represent the only definite constituent of the containing DP, checking definiteness by LDA, as in (49). In such cases, what matters is that the Gen DP is not in DP initial position, i.e., [Spec, DP], being preceded by periphery As, a structure impossible in MR. (49) a. b. din cumplita vră9mă9ie frumos mirositoare a dragostei flori […] a răsări of cruel.the hostility sweet smelling AL love flower will spring “...the sweet smelling flower of love will arise out of that cruel hostility...” însă 9i aceasta pre mai mare a vicle"ugului căptu"ală o făcea yet and this PE bigger AL cheating hiding itCL made …but this she was doing to hide her cheating all the more.” (Cantemir) In the same texts, however, a pre2nominal Gen DP may be in first position and is sufficient to trigger a definite interpretation of the containing DP, just as in English or in MR. This may be interpreted as an instance of LA. As already mentioned above in section 1., LA with the pre2nominal Gen is the only option of MR, examples (50) being syntactically perfect in MR. ALEXANDRA CORNILESCU & ALEXANDRU NICOLAE (50) a. b. 20 …luându'se după a firii sale simŃire (Cantemir) …taking2refl after of his nature feelings… ‘following the feelings if his own nature’ nici vă mieraŃi de ale mele împleticite protase (Cantemir) nor wander about AL my clumsy sentences… ‘Nor do you wonder about my clumsy sentences…. Thus definiteness valuation across a specifier is a general option of OR for all constituents that could, in principle, value the definiteness feature in D. 7. Syntactic ambiguity and the loss of the Lower Definite Article The co2existence of LA and LDA already mentioned above resulted in the occurrence of several systematically ambiguous structures. We briefly discuss two such cases. 7.1 The pre'nominal lexical Genitives As shown in the previous section, the pre2nominal Gen could verify definiteness locally or at a distance, a phenomenon which could lead to interpretative ambiguities. Here is an example: (51) raiul spândzurat cel ce din ["apte ale lumii minuni] unul ieste (Cantemir) paradise.the suspended that which out of seven of the world wonders one is Assuming the DP structure in (25) above and also that the pre2nominal Gen sits in a PossP analogous to the English determiner Gen, example (51) has the following two representations: (52) a. b. DP [2def] Tapte DP [+def] QP Tapte QP Tapte PossP ArtP/NumP ale lumii minuni PossP ArtP/NumP ale lumii[+def] minuni LA LDA Representation (52a) is indefinite, “seven wonders of the world”; representation (52b) is definite, meaning “the seven wonders of the world”. (52b) involves long distance valuation of definiteness across the QP, while in (52a) the cardinal values the feature on D by means of LA. The ambiguity in the data made difficult for the interpreter to establish the type of (in)definiteness checking involved. 7.2 DP'internal inverted predicative adjectives A second type of ambiguity regards the interpretation of pre2nominal As followed by the lower article. In the first place, these are attributive As in a LDA configuration. This is the only interpretation, for instance, after prepositions (53a). On the other hand, the indefinite A may be an emphatic inverted predicative A, originating in a Kaynean ON THE SYNTAX OF ROMANIAN DEFINITE PHRASES 21 small clause, and appearing at the periphery of the DP in a construction devoid of the copula, possibly as in (53b), or across the copula otherwise (53c). (53) a. ...au venit egumenul de BistriŃă cu cinstită cartea mării tale (DÎR) has come abbot.the of BistriŃa with honoured letter.the highness.theGen your “...the Abbot of BistriŃa came with your highness’ honoured letter” b. iar ascunsŭ giudéŃul lui Dumnedzău toate gândurile oamene9ti le strămută (Costin) and hidden judgment.the of God all thoughts.the human themCl moves “and the hidden judgment of God moves all human thoughts” “and the judgment of God, being hidden, moves all human thoughts” c. Iară lunecoasă sunt lucrurile războaielor si în puterea lui Dumnezeu stau. And slippery are things.the of wars and in power of God stand (they) (Costin) “And the matters of wars are slipperyand lie in the power of God.” Ambiguities of this type led to difficulties of establishing the value of the Agree Parameter: Local Agree or Long Distance Agree? Such situations eventually triggered the re2setting of this parameter, in favor of the more constrained grammar, the one in which definiteness is valued by means of LA, a system which generates fewer ambiguities. The re2setting of definiteness valuation illustrates an instance where the Subset Principle is diachronically relevant. Roberts (2007: 260) shows that „the Subset Principle might […] lie behind the phenomenon of ‘restriction of function’, whereby in one system a given operation applies more freely than in another”. 8. Conclusions 1. In Old Romanian, the definite article suffixed to the noun / adjective may occupy the first position of the DP, but also a lower position. In particular, indefinite constituents such as indefinite quantifiers and adjectives or indefinite Gens may precede the definite noun. Old Romanian thus disposes of Long Distance Agree in the valuation of definiteness. Distributionally, the lower article is conditioned mostly by a post2nominal Gen or modifier. 2. The lower article is evidence that the Romanian enclitic definite article originates in a post2posed demonstrative, following the same steps as suggested for Scandinavian by Roehrs (2006). It also confirms that the article should be viewed as a suffix combining with the N in the lexicon. 3. The presence of suffixed definite article leads to a different system of valuing the [idef] feature in D by the [u+def] feature carried by the definite noun. 4. At some point in the evolution of Romanian, definiteness became a concord feature for adjectives; the latter optionally entered the derivation with uninterpretable unvalued definite feature, valued by Agree with the noun. It is reasonable to assume that it was precisely the possibility of valuing definiteness on pre'nominal adjectives that ultimately led to the Modern Romanian requirement that the constituent that values the [idef] feature of D should be the first AP/NP below D. 5. In Old Romanian, Long Distance Agree apparently co2exists with Local Agree, a factor that may cause ambiguity. Consequently, Romanian settles for Local Agree, selecting the more restrictive grammar. ALEXANDRA CORNILESCU & ALEXANDRU NICOLAE 22 6. While in Old Romanian there is Long Distance Agree and Long Distance Move, Modern Romanian loses both options. Changes in these parameters lead to the disappearance of a number of DP structures involving Long Distance Move/Agree. Thus, all patterns involving movement of an XP across the demonstrative or checking of the definiteness feature across a demonstrative are lost. References Akiyama, Masahiro. 2004. “Multiple Nominative Constructions in Japanese and Economy”. Linguistic Inquiry 34.4., 6712683. Barbu, Ana Maria. 2004. Sintaxa determinatorilor. Analiză lingvistică 9i computaŃională. BucureTti: All. Bernstein, Judy. 1991. “DPs in French and Walloon: Evidence for Parametric Variation in Nominal Head Movement”. Probus 3.2, 1012126. Bernstein, Judy. 1993. “The Syntactic Role of Word Markers in Null Nominal Constructions”. Probus 5, 5238. Bernstein, Judy. 2001. “Focusing the “Right” Way in Romance Determiner Phrases”. Probus 13. 1229. Borer, Hagit. 2005. In Name Only. Structuring Sense. Oxford: Oxford University Press. BrâncuT, Gr. 2004. Introducere în istoria limbii române. BucureTti: Editura FundaŃiei România de Mâine. Brugé, Laura. 2000. Categorie funzionali del nome nelle lingue romanze. Milan: Cisalpino. Chomsky, Noam. 1995. The Minimalist Program, Cambridge, Mass : MIT Press. Cinque, Guglielmo. 1993. “A Null Theory of Phrase and Compound Stress”. Linguistic Inquiry 24, 3. Cinque, Guglielmo, 2004. “A Phrasal Movement Analysis of the Romanian DP”. Studia Linguistica et Philologica in honorem D.Irimia ed. by A.Minut & E.Munteanu, IaTi: Ed. Editura UniversităŃii “A. I. Cuza”. 1292142. Cornilescu, Alexandra. 1992. “Remarks on the Determiner System of Rumanian: the Demonstratives AL and CEL”. Probus 4. 1892260. Cornilescu, Alexandra. 1995. “Romanian Genitive Constructions”, Advances in Romanian Linguistics ed. by Guglielmo Cinque & Giuliana Giusti, Oxford: Oxford University Press, 1252. Cornilescu, Alexandra. 2004. “Romanian Genitives Revisited”. Bucharest Working Papers in Linguistics. Coteanu, Ion. 1956. “Despre poziŃia articolului în limba română”. Studii 9i cercetări lingvistice VII, 122. Croitor, Blanca. 2008. “Aspecte privind acordul în determinare în limba română veche”. Limba română − Dinamica limbii, dinamica interpretării ed. by G. Pană Dindelegan, BucureTti: EUB, 2132218. Densusianu, Ovid. 1961. Istoria limbii române. BucureTti: Editura btiinŃifică. Dimitrescu, F. (ed.). 1978. Istoria limbii române. BucureTti: Editura Didactică Ti Pedagogică. Dimitrova2Vulchanova, M. & G. Giusti, 1998. “Fragments of Balkan Nominal Structure”. University of Venice Working Papers in Linguistics 8, 1. 1412172. Dobrovie2Sorin, Carmen. 1987. “A propos de la structure du groupe nominale en roumain”. Rivista di gramatica generativa 12, 1232152. Dobrovie2Sorin, Carmen & Ion Giurgea. 2006. “The Suffixation of Definite Articles in Balkan Languages”. Revue roumaine de linguistique LI, 1, 732104. Farkas, D, H. de Swart. 2007. “Article Choice in Plural Generics”. Lingua 117. 165721676. Farkas, D., K. von Heusinger. 2003. “Stability of Reference and Object Marking in Romanian”. Presentation at Workshop on Direct Reference and Specificity, Vienna, August 2003. GheŃie, Ion. 1975. Baza dialectală a românei literare. BucureTti: Editura Academiei. Giusti, Giuliana. 1993. La sintassi dei determinanti. Padua: Unipress. Giusti, Giuliana. 1996. “Is There a Focus and a Topic Phrase in the Noun Phrase?”. mss., University of Venice. Giusti, Giuliana. 1998. “The Rise of a Functional Category. From Latin ILLE to the Romance Article and Personal Pronoun”. University of Venice Working Papers in Linguistics 8, 2, 53271. Giusti, Giuliana. 2005. “At the Left Periphery of the Romanian Noun Phrase”. On Space and Time in Language ed. by M. Coene & L. Tasmowski, Cluj Napoca: Clusium. 23249. GALR 2005 − Gramatica limbii române. 2005. BucureTti: Editura Academiei Române. Graur, Al. 1967. “De nouveau sur l’article postposé en Roumain”. Revue roumaine de linguistique 1, 3218. Giurgea, Ion Tudor. 2008. Recherches sur la Structure Interne des Pronoms et des Expressions Nominales sans Nom Exprimé. PhD dissertation, Université Paris 7 “Denis Diderot”. Grosu, Al. 1988. “On the Distribution of Genitive Phrases in Romanian”. Linguistics 26, 9312949. Huddlestone, R & G. Pullum. 2002. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Iordan, I. & M. Manoliu. 1965. Introducere în lingvistica romanică. BucureTti: Editura Didactică Ti Pedagogică (Spanish edition − 1972, Manual de lingüística románica, Rev., reelaboración parcial y notas pro Manuel Alvar, Madrid, Editorial Gredos). Isac, Daniela, 2006. “In Defense of a Quantificational Account of Definite DPs”. Linguistic Inquiry 37, 2, 2752288. Julien, Marit. 2005. Nominal Phrases from a Scandinavian Perspective. Amsterdam: John Benjamins. Kamp, H. & U. Reyle. 1993. From Discourse to Logic. Dordrecht: Kluwer. ON THE SYNTAX OF ROMANIAN DEFINITE PHRASES 23 Koptjevskaya2Tamm, Maria. 2005. “Maria’s ring of gold: Adnominal Possession and Non2anchoring Relations in European Languages”. Possessives and Beyond: Semantics and Syntax. ed. by K. Ji2Young et al., Amherst. Laenzlinger, Christopher. 2005. “Some Notes on DP2Internal Movement”. GG@G 4:2272260. Longobardi, Giuseppe. 1994. “Reference and Proper Names: a Theory of N2movement in Syntax and Logical Form”. Linguistic Inquiry 25, 4. 6092665. Longobardi, Giuseppe. 2001. “The Structure of DPs: Some Principles, Parameters and Problems”. The Handbook of Contemporary Syntactic Theory ed. by M. Baltin & C. Collins. Blackwell, 5622603. Lyons, Christopher. 1999. Definiteness, Cambridge: Cambridge University Press. Manoliu2Manea, Maria. 1993. Gramatică, pragmasemantică 9i discurs. BucureTti: Litera. Niculescu, Alexandru, 1990. Outline history of the Romanian Language. Padua: Unipress. Ortmann, A & A. Popescu. 2000. “Romanian definite articles are not clitics”. Clitics in Phonology, Morphology and Syntax ed. by B. Gerlach & J. Grijzenhout. Amsterdam: John Benjamins. Pană Dindelegan, Gabriela. 2008. “Tipuri de gramaticalizare. Pe marginea utilizărilor gramaticalizate ale prepoziŃiilor de Ti la”. Limba română. Dinamica limbii, dinamica interpretării ed. by Gabriela Pană Dindelegan, 2272239. Bucharest, EUB. Pesetsky, D. & E. Torrego. 2007. “The Syntax of Valuation and the Interpretability of Features”. Phrasal and Clausal Architecture: Syntactic derivation and interpretation ed. by S. Karimi, V. Samiian & W. K. Wilkins. Amsterdam: John Benjamins. 2622294. Renzi, Lorenzo. 1993. “L’articolo posposto rumeno in diacronia e in sincronia”. Revue Roumaine de Linguistique XXXVIII, 4, 3072323. Renzi, Lorenzo. 1997. “Fissione di lat. ILLE nelle lingue romanze”. Italica et Romanica. Festschrift für Max Pfister zum 75, ed. by G. Holtus, J. Kramer & W. Sweckhard. Tübingen: Niemeyer. Roberts, Ian. 2007. Diachronic Syntax, Cambridge: Cambridge University Press. Roehrs, Dorian. 2006. The Morpho'Syntax Of The Germanic Noun Phrase: Determiners Move Into The Determiner Prhase. PhD dissertation, Indiana University. Rosetti, Al. 1968. Istoria limbii române de la origini până în secolul al XVII'lea. BucureTti: Editura pentru Literatură Ti Artă. Stateva, Penka, 2002. “Possessive clitics and the structure of nominal expressions”. Lingua 112, 6472690. Stowell, T., 1989. “Subjects, Specifiers, and X2Bar Theory”. Alternative Conceptions of Phrase Structure ed. by M. Baltin & A. Kroch, The University of Chicago Press. Tasmowsky, Liliane. 2009. “Le roumain, entre langues romanes et langues balkaniques”. Conference at the University of Bucharest, may 2009. Zwicky, A & G. Pullum. 1983. “Cliticization vs. Inflection: English n’t”. Language 59, 3, 5022513. Corpus DÎR − Documente 9i însemnări române9ti din secolul al XVI'lea, text stabilit Ti indice de Gheorghe Chivu, Magdalena Georgescu, Magdalena IoniŃă, Alexandru MareT Ti Alexandra Roman2Moraru, Introducere de Alexandru MareT. BucureTti: Editura Academiei, 1979. Costin − Miron Costin, Opere alese, edited by P.P. Panaitescu, BucureTti: Editura Tineretului, 1966 Cronicarii Munteni, edited by Al. Piru, BucureTti, Editura Tineretului, 1964. „LetopiseŃul Cantacuzinesc”, in Cronicarii Munteni. R. Popescu, „LetopiseŃul Bălenilor”, in Cronicarii Munteni. R. Greceanu, „Cronica”, in Cronicarii Munteni. Constantin Cantacuzino, „Istoria łării RumâneTti”, in Cronicarii Munteni. Cantemir − Dimitrie Cantemir, Istoria ieroglifică, ColecŃia „Biblioteca pentru toŃi”. 1983. Editura Minerva, BucureTti. Ureche− Grigore Ureche, LetopiseŃul łărâi Moldovei, ed. by P. P. Panaitescu. 1955. BucureTti: ESPLA. Biblia − Biblia sau dumnezeiasca scriptură, BucureTti, 1688.