The Semantics of Determiners

Maria a citit ziarul.
M. has read newspaper.Def
‘Mary read the newspaper.’

This type of apparent counterexample to both the familiarity and the uniqueness view of definites turns out to be unproblematic given that a natural treatment of the recalcitrant definite DPs here is to see them as referring to types, which are both unique and familiar. (For recent discussion of such DPs as examples of incorporation, see Carlson (2003) and Carlson and Sussman (in press).)

A more problematic case is that of definite DPs referring to non-unique entities often though not always explicitly marked as ‘possessed’ in a possessive relation:

(34) Maria a văzut o pată pe coperta cărţii/ peretele casei.

M. has seen a spot on cover.Def book.Def.Dat/wall.Def house.Def.Dat
‘Mary saw a spot on the cover of the book/the wall of the house.’
Maria s- a dus la plimbare pe malul rîului.
M. Refl. has gone at walk on bank.Def river.Def.Dat
‘Maria went walking on the bank of the river.’

The problem is, of course, that books have two covers, houses have four walls, and rivers have two banks.

The fact that the possessor is definite here is not crucial, as shown by (35):

(35) Maria a văzut o pată pe coperta unei cărţi.

M. has seen a spot on cover.Def a.Dat book.Dat
‘Maria saw a spot on the cover of a book.’

Moreover, being overtly marked as a possessed in a possessor relation is not crucial either, since we have definites in examples like (36):

(36) Paul a deschis geamul.

P. has opened window.Def
‘Paul opened the window.’
Maria a văzut o pată pe perete.¹⁶
M. has seen a spot on wall
‘Maria saw a spot on the wall.’

Obviously, the window Paul opened and the wall with the spot on it were those of the room Paul/Mary were at the time but this is not openly marked in (36). And yet, a definite form is used even though it is not assumed before or after the utterance of these examples that the room in question had only one window or only one wall.

Next, note that although in all these cases we have a definite form in the absence of unique reference, it appears to matter that the number of potential referents is small. This is shown by the contrast between the examples in (34) and the ones in (37):

(37) Maria a văzut o pată pe pagina cărţii /geamul blocului.

M. has seen a spot on page.Def book.Def.Dat window.Def building.Def.Dat
‘Maria saw a spot on the leaf of the book/the window of the building.’

In the absence of context, (37), unlike (34), is interpreted as involving a book with a single page and a building with a single window.¹⁷

A final relevant observation is that there is a sense in which although the referent of the definite DPs in question is not unique in (34) - (36), which referent is chosen from the narrow number of choices available in context is a matter of indifference. Exactly how to state this property or how to capture it is not clear. But there is a sense in which it is a matter of indifference to the speaker (and the context) which cover/window/wall/river bank is involved in these examples even though obviously it may well be that only one was involved and not the other(s).

What does appear to be clear is that the uniqueness side of the requirement associated with definites can be relaxed in certain cases that involve a discourse referent associated with another, familiar referent, under certain conditions. These conditions involve a relatively narrow choice of value for the discourse referent introduced by the definite DP, a choice that is under-determined and irrelevant. Possessive constructions are particularly appropriate for instantiating such cases given that the possessor provides the ‘anchoring’ discourse referent, which explains why they have received particular attention (cf. Barker 2000, 2005, Rawlins 2006.)

The connection between definiteness and being marked as the ‘possessed’ in a possessive construction is morphologically apparent in Romanian. In the rather rare cases when both the possessed and the possessor are indefinite, the ‘possessed’ is marked not only by an indefinite article but also by a post-nominal definite article agreeing with it in gender and number:

(38) Maria a citit o carte a unui politician american.

M. has read a book Def.Fem a.Dat politician american
‘Maria read a book of an American politician.’

It would, however, be a mistake to see this as a property of possessive constructions in isolation. An attractive way of making sense of these cases, suggested in Rawlins (2006) but under different assumptions than our current ones, is to treat them as achieving determined reference by accommodating a discourse referent for the definite DP based on the presence in the discourse of the referent they connect to (the discourse referent introduced by the possessor, if there is one, or, in the absence of a possessor, accommodated given what is known of the discourse). Once such a discourse referent is accommodated, as long as one accommodates a single discourse referent, a definite DP may be used involving that very discourse referent since now determined reference is unproblematic. The issue of differentiating between (37) and (34) boils down to the question of when one can assume the accommodation of a single discourse referent connected to another which does not ensure the uniqueness of the accommodated discourse referent. In the case of the roof of a previously mentioned house, the accommodation of a single roof is unproblematic. But the accommodation of a single wall or window or cover of a building, room or book can be treated as familiar, is problematic given that buildings, rooms and books come with several walls, windows and covers respectively. Apparently, the answer has to do with the fact that the choice is limited, and, moreover, with the fact that which entity one chooses is not relevant to the purposes of the conversation.¹⁸ Obviously, if we have already introduced in the discourse the many windows of a room or the two covers of a book, a subsequent use of a singular definite will not be appropriate any more. Accommodation is a last resort mechanism. In the type of context just mentioned, we already have the relevant referents present and since there are several of them, determined reference cannot be achieved. Note that determined reference is helpful in capturing the affinity between possessed DPs and definiteness: possessed DPs always involve a narrowed down choice of referent in a way that is similar to partitive DPs to be discussed below. Turning these observations into a full-fledged analysis is beyond what we can do here.

So far, we have argued that the ordinary definite -(u)l in Romanian, just like its counterpart, the in English is marked with a feature [DEF] that encodes determined reference. Definite pronouns have determined reference as well but the definite article is incompatible with them presumably because they themselves occur in D. Proper Names too have determined reference. Their inability of occurring with a definite article in Romanian (or English) must be attributed to syntactic considerations. There are languages such as Greek, where personal Proper Names must occur with a definite article, or Italian, where they may, in certain cases.

Staying on the definite side of the determiner tree in Figure 4, the question that arises next is what the difference is between ordinary definite articles and demonstratives such as acest copil/copilul acesta ‘this child’. Going into the details of the syntax of demonstrative DPs in Romanian is not the business of this chapter. Here we will only note that Romanian demonstratives, just like their English counterparts, can be used deictically or not. A typical deictic use is exemplified in (39):

(39) [pointing to a book] Maria a citit cartea aeasta.

M. has read book.Def this
‘Maria read this book.’

Roberts (2002) and Wolter (2006) persuasively argue that English demonstratives do not crucially differ from ordinary definites in that, just like them, they have anaphoric uses and may occur under the scope of universals. An anaphoric use of a demonstrative is exemplified in (40):

(40) Maria a vorbit cu un coleg. Acel coleg a vorbit cu Paul.

M. has spoken with a colleague. That colleague has spoken with P.
‘Maria spoke with a colleague. That colleague spoke with Paul.’

An example where the demonstrative has narrow scope, just like its antecedent, is given in (41) :

(41) Dacă un inspector găseste geamuri sparte, acele geamuri trebuie reparate.

if an inspector finds windows broken those windows must fix.Pl
‘If an inspector finds broken windows, those windows must be fixed.’

We follow Wolter (2006) in taking the demonstrative determiners to encode a restriction concerning the domain where unique reference is achieved. In particular, demonstratives require that the predicative condition introduced by their NP be interpreted relative to a non-default situation, a salient subset of the input DRS. Applying Wolter’s classification for Romanian we get the tree in Figure 6:

definite

unmarked marked

-ul acest/acel

Figure 6: Definite determiner typology

In our terms, the definite article is marked for determined reference, with the feature [DEF]; the demonstrative carries both the [DEF] feature and the special [DEM] feature requiring the interpretation of the NP to be performed relative to a non-default situation. The two determiners are compatible with one another and therefore there is nothing in their semantics that rules out co-occurrence. In fact in Romanian they do indeed co-occur: demonstrative DPs occur either with a demonstrative article only (acest copil ‘this child’) or with a definite article and a postposed demonstrative (copil-ul acesta ‘child-the this’).¹⁹ On the other hand, a demonstrative determiner entails determined reference and therefore it is not surprising to have a language like English that marks determined reference only once in demonstratives and thus does not allow DPs with both the definite article and the demonstrative one. For such languages, both types of determiners introduce a discourse referent (are of type < <e, t>, e >). In Romanian, the demonstrative may function as a determiner, in which case there is no definite article, and then its role is exactly as in English. It may also co-occur with a determiner, in post-nominal position, in which case it functions as an NP modifier imposing a particular condition on the interpretation of the NP. The overall effect in both cases, however, is the same.