MARKING: DATA FROM ROMANCE ALEXANDRU MARDALE
Abstract. It is well known that in certain Romance languages the direct object may be introduced or not by a marker which is homonymous with a corresponding preposition:
pe in Romanian, a in Spanish and Sardinian. This phenomenon is known as Differential Object Marking (DOM).
The aim of this paper is twofold. On the one hand, we present a detailed description of the distribution of DOM in Romance. On the other hand, we sketch a generalization correlating DOM with the denotation of the noun that functions as a direct object.
We begin by briefly examining the parameters that generally determine DOM.
Then we compare Romance marked constructions and describe the similarities and the differences between the three languages presenting this phenomenon. After reviewing a number of previous analyses, we finally adopt a proposal based on the semantic type of the direct object (cf. Cornilescu & Dobrovie-Sorin (2007)). This proposal may be expressed as a twofold generalization: (i) DOM is obligatory only for those objects that are necessarily <e>-type (within the appropriate class of nouns, i.e. nouns with the feature [+ human]); (ii) DOM is excluded with those objects that have a property denotation, i.e. nouns that are <e,t>-type.
1. INTRODUCTION
The aim of this paper is twofold. On the one hand, we present a detailed description of the distribution of Differential Object Marking in Romance (in the rest of this paper, we will use the term DOM to refer to this phenomenon)1. On the other hand, following Cornilescu & Dobrovie-Sorin (2007), we sketch a generalization correlating DOM with the denotation of the noun.
The paper is organised as follows. In section 2 we introduce the data to be discussed. We begin by briefly examining the parameters that generally determine DOM. Then we compare Romance marked constructions and describe the similarities and the differences between the three languages presenting this phenomenon (namely Romanian, Spanish and Sardinian). Section 3 reviews a
1 The Romance phenomenon represents an instance of a more general one known, since Bossong (1985), as Differential Object Marking (on this topic, see Lazard (1982), Karimi (1990, 1996, 1999), Shokouhi & Kipka (2003) for Persian, Rapoport (1995) for Hebrew, Nikolaeva (2001) for Ostyak and Enç (1991), Erteschik-Shir (1997), von Heusinger & Kornflit (2005) for Turkish).
RRL, LIII, 4, p. 449–467, Bucureşti, 2008
number of previous analyses, and section 4 offers a new proposal, based on the semantic type of the direct object. Finally, section 5 provides the conclusions of our research.
2. THE DATA
It is well known that in certain Romance languages the direct object may be introduced by a marker which is homonymous with a corresponding preposition2: pe in Romanian, a in Spanish3 and Sardinian. This section introduces a few examples that briefly illustrate DOM for the three languages:
(a) L-am întâlnit *(pe) Ion. (Romanian)
him-had met PE John
‘I met John.’
(b) Vi *(a) Juan. (Spanish)
saw A John
‘I saw John’
(c) An furatu *(a) Ercole. (Sardinian)
have stolen A hercules
‘They have stolen Hercules.’
In these examples, DOM is obligatory, but it may also be optional or excluded depending on a variety of parameters (see data and tables in section 2.2).
The next paragraphs are dedicated to the examination of these parameters.
2.1. The parameters
As shown by several authors (see, among others, Bossong (1985), Lazard (1994), Laca (1995, 2001), von Heusinger & Kaiser (2005)), there are three main parameters that determine DOM cross-linguistically: (i) animacy, (ii) referentiality and (iii) topicality.
According to Laca (2006), parameters (i) and (ii) are said to be local parameters, i.e. they concern the inherent properties of the direct object, while parameter (iii) may be understood as a global parameter, i.e. it concerns the context in which the object occurs.
2 Because of this homonymy, the phenomenon is also called prepositional accusative.
3 In this paper, we analyse data from European Spanish. It is important to point out this aspect since there is a strong variation between the latter and the dialects spoken in Latin America.
2.1.1. Global parameter(s)
Topicality may interact with – and most of the time may be triggered by – additional factors such as the lexical nature of the verb, modification, anaphoricity, clitic doubling, prosody and / or accentuation, preverbal position, information structure etc.
From this perspective, it can be observed that objects which do not trigger DOM (because they don’t have one of the relevant properties in (i) or (ii)), can be marked if they are on the incidence of an additional factor. In other words, global parameters may cancel local ones and thus objects that due to their local properties would not be introduced by the marker might allow it depending on the context.
Such a situation arises for the following Spanish example (compare (d) and (e)):
(d) (A) la sacristiai lai traspasaba un buen sablazo de sol.
‘A good cut of sun pierced the vestry.’
(e) Un buen sablazo de sol traspasaba (*a) la sacristia.
A quick examination of this construction shows that, in spite of its inanimate features (cf. parameter (i)), the direct object la sacristia ‘the vestry’ may be marked by a (as in (d)) due to its left dislocated position (and maybe to a special accentuation or to clitic doubling). Moreover, if the direct object stays in its in situ position, i.e. the post verbal position (as in (e)), it does not trigger marking altough we would expect it since the noun has a referential / specific reading (cf. parameter (ii)).
2.1.2. Local parameters
Going back to the so-called local parameters, they have been described in terms of scales formed by different values (see Comrie (1975), Croft (1988) and the already cited authors). For instance, the following values are associated to animacy: human > animate > inanimate. Referentiality, which is often associated to definiteness (see, among others, von Heusinger & Kaiser (2003, 2005), Leonetti (2003)), forms a scale composed by the following values: definite > indefinite specific > indefinite non-specific.
More recently, Aissen (2003) and Laca (2001, 2006) propose a refined version of these scales. These authors take into consideration, on the one hand, the (grammatical) category of the direct object, and, on the other hand, they combine the various values of the two scales. Consequently, they propose the following improved version:
(A) human pronouns > human proper nouns / animate pronouns (B) human definite NPs / animate proper names / inanimate pronouns
(C) human indefinite specific NPs > animate definite NPs > inanimate proper nouns
(D) human non specific NPs > animate indefinite specific NPs > inanimate definite NPs
(E) animate non specific NPs > inanimate indefinite specific NPs > inanimate non specific NPs
With these considerations in mind, the general understanding of DOM may be expressed as follows (apud von Heusinger & Kaiser (2005: 38)):
A high position on a scale tends to trigger DOM and a low position tends to block DOM.
We adopt these scales with some minor modifications concerning their presentation.
2.2. The distribution of DOM in Romance
In this section, we offer a detailed description of DOM in Romance. On the basis of the scales given in (A) – (E), our goal is to point out the similarities and the differences between the three languages.
As we mentioned above, the three Romance languages have in common the fact that they are generally sensitive to the inherent properties of the direct object.
In some cases, they can be sensitive to contextual factors (see the Spanish example in (d)). However, the three languages differ with respect to the value of the scale which triggers DOM (see data above).
2.2.1. Romanian4
The marker pe is obligatory with direct objects realised as:
I. Specific pronouns: strong personal pronouns [+ human] (1), deictic and anaphoric pronouns [+/- animate] (2)
(1) Li-ai invitat *(pe) eli. ‘You invited him.’
(2) a. Li-am văzut *(pe) {acestai / cel verdei / al meui}.
‘I saw this one / the green one / mine.’
b. Am văzut filmuli *(pe) carei mi li-ai recomandat.
‘I saw the movie that you recommended to me.’
II. Specific NPs: proper nouns [+/− human] (3), definite NPs [+ human] (4), indefinite specific NPs [+ human] (5)
4 Romanian data are my own examples; some of them are adapted from Niculescu (1965), Cornilescu (2000), Farkas & von Heusinger (2003).
(3) Li-am chemat *(pe) Lupui.
‘I called Lupu.’
(4) a. Am văzut-oi *(pe) mamai ta.
‘I saw your mother’
b. Am căutat-oi *(pe) studentai din prima bancă.
‘I looked for the student in the first row.’
(5) Îl caut *(pe) un student (care ştie engleză). (compare with (10a)) ‘I am looking for a student (who speaksIND English).’
III. Bare Quantifiers [+ human]:
(6) a. N-am văzut *(pe) nimeni.
‘I didn’t see anyone.’
b. Ai văzut *(pe) {cineva / cine / oricine / fiecare} la petrecere ? ‘Have you seen somebody / anybody / everybody at the party ?’
The marker pe is excluded with:
I. Specific NPs: proper nouns [– animate] (7), definite NPs [– human] (8), indefinite specific NPs [− human] (9):
(7) Am văzut / vizitat (*pe) Napoli.
‘I saw / visited Naples.’
(8) a. Am chemat (*pe) pisica vecinului.
‘I called the neighbour’s cat.’
b. Am văzut (*pe) filmul despre care mi-ai vorbit.
‘I saw the movie you told me about.’
(9) Am văzut (*pe) un cal care are o pată albă pe frunte.
‘I saw a horse with a white spot on its forehead.’
II. Non specific NPs [+/– human]:
(10) a. Caut (*pe) un student (care să ştie engleză). (compare with (5)) ‘I am looking for a student (who speaksSUBJ English).’
b. Caut (*pe) secretară / reviste.
‘I am looking a secretary / some magazines.’
c. Am întâlnit (*pe) câteva persoane interesante azi.
‘I met some interesting people today’.
d. Am citit (*pe) multe cărţi despre subiect.
‘I have read a lot of books on this topic.’
III. Bare Quantifiers [− human]:
(11) a. N-am văzut (*pe) nimic.
‘I saw nothing’
b. (*Pe) {ce / ceva / orice} ai citit ?
‘What have you read ? Have you read something / anything ?’
IV. Generic NPs [+/– human]:
(12) a. Ion adoră / respectă (*pe) fetele inteligente.
‘John loves / respects intelligent girls.’
b. Ion adoră (*pe) câinii cu labele albe.
‘John loves dogs with white paws.’
c. Ion adoră (*pe) romanele poliţiste.
‘John loves crime stories.’
V. Phrases where the subject is interpreted as Agent or Cause:5 (13) Acizii atacă (*pe) metale.
‘Acids attack metals.’
VI. Left dislocated NPs [– human]6: (14) a. (*Pe) câine, l-am numit Lupu.
‘I called the dog Lupu.’
b. (*Pe) vapor, l-am admirat de două ori.
‘I admired the ship twice.’
2.2.2. Spanish7
DOM in Spanish is less straightforward than in Romanian. Apart from the inherent properties of the direct object, various other parameters may play a role for DOM. More precisely, the verb type (and the interpretation of its subject – see Torrego Salcedo (1999)), as well as the position of the object may be DOM triggers.
5 Following Torrego Salcedo (1999) this context is called the Agentivity Constraint : if the direct object triggers marking, the subject must be interpreted as an Agent or a Cause. See below a few relevant examples for Spanish (cf. (21)).
6 An apparent exception to this ‘rule’ is the following example : *(Pe) trandafiri li-a lăsat albina la urmă ‘The bee left the rose at the end.’ The presence of the marker with this inanimate direct object is not due to the preverbal position, but rather to the fact that the rose has human attributes (notice that the context from which this example is taken is a fairytale). In other words, we are dealing with a case of personification and it is for this reason that the marker is allowed.
7 Spanish data are adapted from Roegiest (1980), Lois (1982), Laca (1995, 2006), Torrego Salcedo (1999), von Heusinger & Kaiser (2005).
The marker a is obligatory with:
I. Specific pronouns [+ human]: strong personal pronouns (15), deictic and anaphoric pronouns (16)
(15) {Lo / la} arrestaron *(a) {él / ella}.
‘They arrested him / her.’
(16) Escucho *(a)l mío.
‘I am listening to mine.’
II. Specific NPs: proper nouns [+/- human] (17), definite NPs [+ human] (18), indefinite specific NPs [+ human] (19)
(17) a. Pedro mató *(a) Juan.
‘John killed Mary.’
b. Vi *(a) Dorotea.
‘I saw Dorotea.’
(18) a. Juan insulta *(a) este señor.
‘John is insulting this man.’
b. No hemos visto *(a) su padre.
‘We didn’t see his father.’
(19) a. Busco *(a) una cocinera (que sabe inglés). (compare with (24a)) ‘I am looking for a cook (who speaksIND English).’
b. Besaron *(a) un niño en la frente. (compare with (24b)) ‘They kissed a child on the forehead.’
c. Besaron *(a) un niño en un segundo. (compare with (24b)) ‘They kissed a child in one second.’
III. Bare Quantifiers [+ human]:
(20) a. Escucho *(a) {alguien / todos}.
‘I am hearing / listening to somebody / everybody.’
b. i. No escucho *(a) nadie.
‘I am not hearing / listening to anybody.’
ii. No los escucho *(a) ninguno de los dos.
‘I am not listening to any of them.’
IV. Phrases where the subject is interpreted as Agent or Cause:
(21) a. Este abogado escondió *(a) muchos prisioneros.
‘This lawyer has hidden many prisoners.’
b. La esperanza sostiene *(a) la voluntad.
‘Hope strengthens will.’
c. Preceden *(a) cada fragmento unas notas.
‘(A few) notes precede each fragment.’
d. Los ácidos atacan *(a) los metales.
‘Acids attack metals.’
The marker a is excluded with:
I. Specific NPs: proper names [− animate] (22), definite NPs [− human] (23), indefinite specific NPs [− humain] (24)
(22) Vi (*a)l Titanic.
‘I saw the Titanic.’
(23) Compré (*a) esta casa.
‘I bought this house’.
(24) a. Busco (*a) un perro con patas blancas.
‘I am looking for a dog with white legs.’
b. Busco (*a) un CD de Domingo.
‘I am looking for a CD of Domingo.’
II. Non specific NPs [+/− human]:
(25) a. Busco (*a) una cocinera (que sepa inglés). (compare with (19a)) ‘I am looking for a cook (who knowsSUBJ English).’
b. Besaron (*a) un niño. (compare with (19a-b)) ‘They kissed a child.’
c. Se contratan (*a) meseros (serios) / secretarias (rubias).
‘They are hiring a (serious) waiters / (blonde) secretaries.’
III. Bare Quantifiers [− human]:
(26) No he visto (*a) nada.
‘I saw nothing’
IV. Generic NPs [+/− human]:
(27) a. Juan adora / respeta (*a) las chicas inteligentes.
‘John loves / respects intelligent girls.’
b. Juan adora (*a) los perros con patas blancas.
‘John loves dogs with white paws.’
c. Juan adora (*a) las novelas policiales.
‘John loves crime stories.’
Finally, the marker a is optional with:
Left dislocated NPs [+/− animate]:
(28) (A) la sacristiai lai traspasaba un buen sablazo de sol.
‘A good cut of sun pierced the vestry.’
2.2.3. Sardinian8
Unlike Romanian and Spanish, Sardinian is not systematically sensitive to the inherent properties of the direct object. This is to say that objects that are on a high position both on the animacy and definiteness scale may not (or may optionally) trigger DOM (see the examples at the end of this section).
The marker a is obligatory with:
I. Specific pronouns [+ human]: strong personal pronouns (29), deictic and anaphoric pronouns (30)
(29) a. Appo vistu solu *(a) isse.
‘I saw only him.’
b. Non connosco *(a) issa.
‘I do not know her.’
(30) *(A) kie kilkasa?
‘Whom are you looking for?’
II. Specific NPs: proper nouns [+/− animate] (31), certain definite NPs [+ human] (32)
(31) a. Unu cazzadore at moltu *(a) Kira.
‘A hunter killed Kira.’
b. Appo vistu *(a) Napoli.
‘I saw Naples.’
8 The Sardinian data are due to Bossong (1985), Jones (1995), Floricic (2003), Mensching (2005), Molinu (p.c.).
(32) Appo vistu *(a) frate tuo / *(a) babbu / *(a) duttore Ledda.
‘I saw your brother / grandfather / doctor Ledda.’
III. Bare Quantifiers [+ human]:
(33) a. Appo invitadu *(a) tottu cantos.
‘I invited everybody.’
b. No appo bidu *(a) nesciune.
‘I saw nobody.’
The marker a is excluded with:
I. Deictic and anaphoric pronouns [− human]:
(34) (*A) cale keres comporare?
‘Which one do you want to buy?’
II. Specific NPs [- human]: definite NPs (35), indefinite specific NPs (36) (35) b. Appo vistu (*a) su cane.
‘I saw your dog.’
c. Appo vistu (*a) sa makkina.
‘I saw your car.’
(36) So kilkende (*a) una makkina ki appo acciappadu custu mandzanu.
‘I am looking for a car that I have seen this morning.’
III. Non specific NPs [+/− human]:
(37) Appo vistu (*a) unu pastore / (*a) (metas) sordatos / (*a) una makkina.
‘I have seen a shepherd / (a lot of) soldiers / a car.’
IV. Bare Quantifiers [− humain]:
(38) No appo vistu (*a) nudda.
‘I saw nothing.’
V. Generic NPs [+ human]:
(39) a. Juanne istimat / rispetat (*a) sos istudiantes intelligentes.
‘John loves / respects intelligent students.’
b. Juanne istimat (*a) sos canes de cursa / (*a) sos filmis d’atzione.
‘John loves race dogs / action movies.’
VI. Phrases where the subject is interpreted as Agent or Cause:
(40) Sos atzidos attacan (*a) sos ferros.
‘Acids attack metals.’
The marker is optional with:
I. Certain specific NPs [+ human] : definite NPs (41), indefinite specific NPs (42) (41) a. An assessinatu (a) su re.
‘They have assassinated the King.’
b. Maria at vistu (a) su dottore.
‘Maria saw her doctor.’
(42) So kilkende (a) unu professore ki appo acciappadu custu mandzanu.
‘I am looking for a teacher that I have seen this morning.’
II. Left dislocated NPs [+ human]:
(43) (A) su preiderui, invitadu li’as a su matrimoniu?
‘The priest, did you invite him at the wedding ?’
2.3. Summary of the empirical results
The following tables summarize our findings so far:
Table 1
Specific pronouns Specific NPs
strong personal pronouns
deictic &
anaphoric pronouns
proper names definite NPs indefinite specific NPs hu
ma n
ani ma te
ina nim ate
hu m an
anim ate
ina nim ate
hu m an
ani mat e
inan imat e
hum an
ani ma te
inan imat e
hu ma n
in/a nim ate
R + Ø Ø + + + + + - + - - + -
S p
+ Ø Ø + + - + + - + - - + -
S a
+ Ø Ø + - - + + + ± - - ± -
Table 2 Non
specific NPs
Bare Quantifiers Generic NPs Topicalized NPs
human / in/animate
human in/animate human in/animate preverbal hum- in/anim.
agentivity constraint R - + - - - - - Sp - + - - - + + Sa - + - - - ± -
2.4. Interpretation of the tables
Considering the data we have presented in the previous subsections, a few points need to be made.
The first one is that DOM in Romance confirms the predictions of the scales given earlier. Put differently, the higher a direct object is on the relevant scale, the more likely it is to be marked.
The second and most interesting point concerns (micro)variation within the Romance family. From this perspective, we can observe two types of variation:
(i) a strong variation (see the yellow cells of the tables) which could be phrased as follows:
(a) Spanish and Sardinian are sensitive to the animacy parameter when the direct object is realised as a deictic pronoun;
(b) Romanian and Spanish are sensitive to the animacy parameter when the direct object is realised as a proper noun;
(c) Spanish is sensitive to the verb type – i.e. the agentivity constraint (see Torrego Salcedo (1999) –, while Romanian and Sardinian are not.
(ii) a weak variation (see the blue cells of the tables) which could be phrased as follows:
(a) Sardinian does not systematically mark human specific direct objects, while Romanian and Spanish do;
(b) Romanian and Sardinian seem to be less sensitive to the left dislocated position than Spanish.
3. PREVIOUS ANALYSES
The contrasts observed so far have been the topic of numerous and controversial analyses. In this section, we briefly review a number of these analyses, which basically pertain to three approaches.
3.1. Functionalist hypothesis: distinguishing between subject and direct object9
The main argument of this type of analyses is based on the possible
‘confusion’ between subject and object. In languages with a relatively free word order (such as Romanian and Spanish), these markers are used to distinguish between subject and direct object, especially when the latter has the properties of a typical suject.
More precisely, this kind of situation may be encountered when the subject, the direct object and the verb formally refer to a third person, as in (44a) and (45a):
(44) a. TatălS/DO iubeşte copilulDO/S. (Romanian) ‘The father loves the child.’
b. *(Pe) tatăDOi îli iubeşte copilulS. c. TatălS îli iubeşte *(pe) copilDOi.
(45) a. Perseguía el guardiaS/DO el ladrónDO/S. (Spanish) ‘The guardian followed the burglar.’
b. Perseguía al guardiaDO el ladrónS. c. Perseguía el guardiaS al ladrónDO.
Consequently, the insertion of the marker allows identification of the direct object (compare (44b) with (44c) and (45b) with (45c)).
This explanation presents the advantage of applying to cases where the direct object has a human referent, as well as to those in which the referent is inanimate:
(46) Cui *(pe) cui se scoate. (Romanian)
‘A nail takes out another nail.’
(47) Los ácidos atacan (a) los metales. (Spanish)
‘Acids attack metals.’
However, it should be noted that there are various cases where there cannot be such an ambiguity, i.e. when the subject is null, although the marker is obligatory:
(48) (Îl) văd *(pe) Ion / el. (Romanian)
‘I (can) see John / him.’
9 See Diez (1876), Hills (1920), Lenz (1920, 1944), Puşcariu (1922), Meier (1947, 1948), Onu (1959), Niculescu (1959), Gramatica Academei (1963, 2005), Guţu Romalo (1973), RAE (1973), Pană Dindelegan (1997, 1999).
(49) Veo *(a) Juan / él. (Spanish) ‘I (can) see John / him.’
3.2. Lexical semantic hypothesis: a and pe as markers of ‘personal gender’10
The expression personal gender coming from traditional grammar refers to nouns with the feature [+ human]. The main argument of this type of analyses is that DOM in Romance generally appears with direct objects denoting humans or animate entities.
As we already saw in the previous section, there are at least three exceptions to this rule:
(i) in Romanian (and sometimes in Spanish), certain marked pronouns do not systematically refer to humans. In the following examples, the demonstrative acesta ‘this one’ and the possessive el mío ‘mine’ may refer to an inanimate entity (a CD, for instance):
(50) Îl voi lua *(pe) acesta. (Romanian)
‘I will take this one.’
(51) Escucho (a)l mío. (Spanish)
‘I am listening to mine.’
(ii) marked proper nouns do not always have a human referent. In this respect, note the special case of Sardinian which marks all proper names regardless of their referent:
(53) Appo vistu *(a) Napoli. (Sardinian)
‘I saw Naples.’
(iii) finally, it is not true that all nouns referring to humans (even when specific) trigger DOM:
(54) a. An saludadu cussu síndigu. (Sardinian)
‘They greeted that mayor.’
b. Appo biu is óminis de s’atra dí.
‘I saw the men of the other day.’
10 See Spitzer (1928), Racoviţă (1940), Hatcher (1942), Graur (1945), Gramatica Academiei (1963, 2005), Niculescu (1959, 1965), Pană Dindelegan (1997), Cornilescu (2000).
3.3. Multiple parameters11
As we have seen in the previous sections (see more precisely § 2.1), this phenomenon cannot easily be explained in terms of dependence on a single parameter.
In other words, it cannot be reduced to a simple dichotomy represented by the presence or absence of a given feature.
In fact, most analyses agree on the fact that there is a hierarchy of parameters that intervene in DOM (recall the distinction between local and global parameters).
This hierarchy allows us to make predictions about which objects may be marked.
We therefore agree with Lazard (1994) and Aissen (2003) who clearly expressed these ideas:
(i) ‘la marque […] se trouve toujours du côté du plus défini / plus humain et son absence du côté du moins défini / moins humain. Le marquage de l’objet est donc corrélatif de son individuation: plus l'objet est fortement individué, plus il a des chances d'être marqué comme tel’ (Lazard (1994 : 230))
(ii) ‘the higher in prominence a direct object, the more likely it is to be overtly case-maked’ (Aissen (2003 : 2))
Another important aspect to point out with respect to DOM in Romance is that the parameters governing this phenomenon should not be considered individually. In order to get an appropriate description and explanation of DOM we need to combine all factors we mentioned. As we will see in the next section, each of them represents a necessary but not sufficient condition for marking.
4. A GENERALIZATION IN TERMS OF SEMANTIC TYPES According to our observations so far, it is clear that the (Accusative) markers pe and a, respectively, are more than an alternative case-marking strategy (see also Cornilescu & Dobrovie-Sorin (2007), footnote 15).
This can be easily proved by examining the following constructions:
(55) a. Întotdeauna m-a invitat pe mine. (Romanian)
‘(S)he always invited meACC.’
b. *Întotdeauna m-a invitat pe eu.
‘(S)he always invited meNOM.’
(56) Compré esta casa. (Spanish)
‘I bought this house’.
11 See Bossong (1985, 1998), Croft (1988), Lazard (1994), Aissen (2003), von Heusinger &
Kaiser (2005), Laca (2006).
In the Romanian example given in (55a), the direct object is (morphologically) marked with the Accusative case. The personal pronoun mine
‘me’ is an Accusative form as opposed to the Nominative form eu ‘I’ in (55b).
Since case is already marked on the pronoun in (55a), we may ask why the marker pe is still necessary. Obviously, some link must exist between pe and the Accusative, since the former is not compatible with the Nominative (see (55b)).
However, this is not reason enough to consider pe as a case-marker.
Moreover, the Spanish example in (56) shows that the object esta casa ‘this house’ may appear without being marked. According to current case-marking theories (see, among others, Chomsky (1981, 1995)), a direct object receives Accusative from its V-sister. This means that the DP under discussion has already checked case and, thus, can appear without a.
In what follows, we will not try to establish the precise categorial status of these markers12: we shall rather focus on the role they play when they precede a direct object.
We would like to propose an alternative way13 to express the conditions under which they may be used.
In this respect, let us recall the analyses of Kamp (1981), Heim (1982) and, more recently, the ones of Kleiber & al. (2001) and Dobrovie-Sorin & Beyssade (2005). According to these authors, nominal expressions may be analysed as having three types of denotation: (i) expressions denoting individuals, i.e. <e>- type expressions; (ii) expressions denoting properties, i.e. <e,t>-type expressions;
(iii) expressions denoting generalized quantifiers, i.e. <<e,t>,t>-type expressions.
For the present purposes, we assume this general classification without discussing it.
Going back to what we have observed in section § 2. above, it may be interesting to correlate DOM with the semantic type of the direct object. To express this correlation, we propose the following twofold generalization:
In Romanian, (European) Spanish and Sardinian:
(A) the markers pe and a are excluded with those direct objects that denote properties (type <e,t>)
(B) the markers pe and a may appear only with direct objects referring to specific individuals (type <e>) or with universal quantifiers (type <<e,t>,t>)
The first part of this generalization (A) explains why non specific nouns and bare nouns do not trigger DOM in any of the three languages (see (12), (27), (39), on the one hand, and (10), (25), (37), on the other hand).
12 The reader is referred to Mardale (2007) for a discussion on this topic.
13 In line with McNally (1992, 1995), Dobrovie-Sorin (1997), Dobrovie-Sorin & Laca (2003), Farkas & de Swart (2003), Leonetti (2003), Bleam (2004), van Geehoven & McNally (2005).
The second part of the generalization (B) explains why all three languages mark: (i) strong personal pronouns (see (1), (15) and (29)); (ii) deictic and anaphoric pronouns (when they have a human referent) (see (2), (16) and (30));
(iii) specific definite NPs and indefinite specific NPs when they have a human referent (3) – (5), (17) – (19) and (31) – (32), (41) – (42)); (iv) bare quantifiers when they have a human referent (see (6), (20) and (33)); (v) preverbal ODs (see (28) and (43)).
The second part of the generalization must however be considered as a first necessary but not sufficient condition for DOM since there are certain <e> type objects that do not allow the markers (see (7) – (9), (22) – (24) and (34) – (36)) because they do not belong to the appropriate class of nouns (i.e., those marked [+
human]).
In that case, the denotation of the noun does not trigger DOM and must be corroborated with the (human or animate) nature of its referent. The latter remark must therefore be understood as a second necessary condition for DOM. This condition is fulfilled differently by those languages depending on their sensitivity to the animacy parameter, parameter which does vary, as we have seen in section
§ 2.3.
Contrary to what happens in Romanian and Sardinian, DOM in Spanish may be subject to a third necessary condition represented by certain properties of the verb – see again Torrego Salcedo’s (1999) Agentivity Constraint (examples (13) and (40) vs. (21)).
Obviously, the generalization we offer does not embrace the hole complexity of the phenomenon. This was not our goal in this section. We only wished to explore another way of explaining DOM. This approach is in line with the multiple factors analysis which we refferred to above.
Besides, much remains to do if we want this generalization to fit with the complexity of DOM. It also raises a number of issues which require further investigations. For instance, if we admit that DOM is excluded with property- denoting nouns, can we say that these contexts must be analysed as semantically incorporating structures? Conversely, is it possible that the presence of the markers in question block semantic incorporation? We leave these questions for further research.
5. CONCLUDING REMARKS
In this paper, we have tried to examine the distribution of DOM in the Romance language family from a comparative point of view. We have concentrated on the three languages that show a broad use of DOM: Romanian, (European) Spanish and Sardinian.
Following Aissen (2003), Laca (2006), we have seen that it is reasonable to adopt an analysis in termes of scales or hierarchy of parameters. According to these scales, we have observed that DOM in Romance generally depends on their highest values. Beside this common behaviour, we have also observed that DOM in Romance presents a number of strong and weak variation points.
The strong variation concerns the following aspects: (i) Spanish and Sardinian trigger DOM in a different way than Romanian when the direct object is realised as a deictic or anaphoric pronoun. The former are more sensitive to the (human) nature of the entity to which the pronoun refers, while the latter is not; (ii) In addition, Sardinian is different from Romanian and Spanish regarding the referent of direct objects when the latter are realised as proper nouns; (iii) Spanish is sensitive to an external-object constraint (namely the verbal type), while Romanian and Sardinian are not.
The weak variation may be expressed as follows: (i) Romanian and Sardinian are less sensitive than Spanish with respect to the preverbal (i.e. topicalized) position; (ii) Within direct objects that are not realised as pronouns or proper nouns, Sardinian seems to be less sensitive than Romanian and Spanish to the animacy and the definiteness scales.
On this basis, we then sketched a generalization of DOM use in Romance.
Our generalization is based on the semantic type of the object and may be expressed as a twofold hypothesis: (i) Marking may appear only with those objects that are necessarily <e>-type (within the appropriate class of nouns, i.e. nouns with the feature [+ human]); (ii) Marking is excluded with those objects that have a property denotation, i.e. nouns that are <e,t>-type.
SELECTED REFERENCES
Aissen, J., 2003, “Differential Object Marking: Iconicity vs. Economy”, Natural Language and Linguistic Theory, 21, 435−483.
Bleam, T., 2004, “A Property Analysis of Weak Nominals in Spanish: Bare Nominals and Prepositionless Accusatives”, Ms. Université Paris 7.
Bossong, G., 1985, Empirische Universalienforschung. Differentielle Objektmarkierung in den neuiranischen Sparchen, Tübingen, Narr.
Cornilescu, A., 2000, “Notes on the Interpretation of the Prepositional Accusative in Romanian”, Bucharest Working Papers in Linguistics, 1, 91−106.
Cornilescu, A., C. Dobrovie-Sorin (to appear), “Clitic Doubling. Complex Heads and Interarboreal Operation’, Ms. University of Bucharest / Paris 7.
Dobrovie-Sorin, C., 1994, The Syntax of Romanian, Berlin, Mouton de Gruyter.
Dobrovie-Sorin, C., 1997, “Classes de prédicats, distribution des indéfinis et la distinction thétique- catégorique”, Le gré des langues, 20, 58−97.
Dobrovie-Sorin, C., B. Laca, 2003, “Les noms sans déterminant dans les langues romanes”, in:
D. Godard (ed.), Les langues romanes. Problèmes de la phrase simple, Paris, Editions du CNRS, 235−281.
Farkas, D., K. von Heusinger, 2003. “Stability of Reference and Object Marking in Romanian”, Ms.
Universität Stuttgart.
Fish, G., 1967, “A with Spanish Direct Object”, Hispania, 50, 1 & sq.
Floricic, F., 2003, “Notes sur l‘accusatif prépositionnel‚ en sarde”, Bulletin de la Société de Linguistique de Paris, 98, 247−303.
von Heusinger, K., G. A. Kaiser, 2005, “The evolution of differential object marking in Spanish”, Proceedings of the Workshop: Specificity and the Evolution / Emergence of Nominal Determination Systems in Romance, U. Konstanz, 33−69.
Ionescu, E., 2000, “The Role of pe in the Direct Object construction in Romanian (some critical remarks)”, Bucharest Working Papers in Linguistics, 1, 81−91.
Laca, B., 1995, “Sobre el uso del acusativo preposicional en español”, in: C. Pensado (ed.), El complemento directo preposicional, Madrid, Visor Libros, 61−91.
Laca, B., 2006, “El objeto directo”, in: C. Company (ed.), Sintaxis historica del español. Vol 1: La frase verbal, México, Universidad Nacional de México.
Lazard, G., 1994, L’Actance, Paris, Presses Universitaires de France.
Leonetti, M., 2003, “Specificity and Differential Object Marking in Spanish”, Catalan Journal of Linguistics, 3, 75−114.
Lois, X., 1982, Sur l’Accusatif prépositionnel, Mémoire de maîtrise, Université Paris 8.
Lujan, M., 1977, “Direct Object Nouns and the Preposition a in Spanish”, Texas Linguistic Forum, 10.
Mardale, A., 2007, Les prépositions fonctionnelles du roumain: étude comparative, PhD Dissertation, Université Paris 7 & Universitatea din Bucureşti.
Menshing, G., 2005, “Remarks on Specificity and Related Categories in Sardinian”, Proceedings of the Workshop: Specificity and the Evolution / Emergence of Nominal Determination Systems in Romance, U. Konstanz, 81−106.
Niculescu, A., 1959, “Sur l’objet direct prépositionnel dans les langues romanes”, Recueil d’études romanes : IXème Congrès International de linguistique romane à Lisbonne.
Onu, L., 1959, “L’origine de l’accusatif roumain avec p(r)e”, Recueil d’études romanes, IXème Congrès International de linguistique romane à Lisbonne, Bucarest.
Pană Dindelegan, G., 1997, “Din nou despre statutul prepoziţiei. Cu referire specială la prepoziţia PE”, Limba română, 1−3, 165−174.
Pensado, C., 1985, “La creacion del Objeto directo preposicional y la flexion de los Pronombres Personales en las lenguas romanicas”, Revue Roumaine de Linguistique, 30, 123−158.
Pensado, C., 1995, El complemento directo preposicional, Madrid, Visor Libros.
Puşcariu, S., 1922, “Despre p(r)e la acuzativ”, Dacoromania, 2, 565−581.
Roegiest, E., 1979, “Autour de l’accusatif prépositionnel dans les langues romanes”, Vox Romanica, 38, 37−54.
Rohlfs, G., 1971, “Autour de l’Accusatif prépositionnel dans les langues romanes (concordances et discordances)”, Revue de linguistique romane, 35, 312−334.
Spitzer, L., 1928, “Rum. P(r)e, Span. a vor persönalichem Akkusativobject”, ArPh, 48, 423−432.
Torrego Salcedo, E., 1999, “El complemento directo preposicional”, in: I. Bosque Muñoz, V. Demonte Barreto (eds.), Gramatica descriptiva de la lengua española, Madrid, Espasa- Calpe, 1779−1807.
