The demonstrative determiners are ce (cet) M.SG, cetteF.SG and cesM/F.PL: (9) a. ce garçon b. cette fille c. ces enfants
DEM.M.SG boy DEM.F.SG girl DEM.M/F.PL children ‘the/this/that boy’ ‘the/this/that girl’ ‘the/these/those children’
Used on their own, demonstrative determiners have weak deictic force, only, and sometimes function as mere definiteness markers (see the translations of the examples in (9)). Note that they fail to distinguish degrees of proximity to any participant in the discourse (again, see the translations in (9)). Also, unlike true deictic determiners, they can introduce discourse-novel referents, as in (10):
(10) Il y a ce type qui me harcèle.
it there has this bloke who me bothers ‘There’s a bloke hassling me.’
Finally, unlike deictic determiners, demonstrative determiners in French are compatible with restrictive relatives. Compare (11a, b):
(11) a. *Je n’ai lu que ce livre-là que tu m’as donné hier.5 (deictic reading) I NEG-have read but this book-there that you me-have given yesterday
b. Je n’ai lu que ce livre que tu m’as donné hier. (non-deictic reading only) I NEG-have read but this book that you me-have given yesterday
In fact, the force of demonstrative determiners is so weak that Gary-Prieur (2001) claims that they can even introduce 6
generic DP*s, provided the generic referent is being contrasted with some other generic referent(s).
The reinforcers -là/-ci are weak forms of the distal/proximal locative adverbials ici ‘here’ and là ‘there’. The full forms 7
occur in ici/là-bas ‘down (t)here’ and ici/là-dedans ‘in (t)here’. The weak forms occur within the demonstrative mass and count proforms cela/ceci ‘that/this’ and celui/celle(s)/ceux-là/ci ‘that/this/those/these one(s)’ and the presentative voilà/ voici. Speakers in sub-Saharan Africa and Southern France use -là even with the definite article: l’autre-là ‘the other one’, le Boeing-là ‘the Boeing’ (Queffélec 2000b: 827). Distal -là is encroaching on proximal -ci (Goosse 2000: 130). Leeman- Bouix (1994: 140–3) attributes the loss of the -là/-ci contrast, at least in part, to the processing load associated with centre embedding (Dean Fodor 1995). In (i) the hearer has to associate celui-ci with Marc and celui-là with Jean, which disrespects linear order:
(i) J’ai vu Jean et Marc; celui-ci a aimé mon projet, celui-là le trouvait mauvais. I-have seen J. and M.; this-one has liked my plan, that-one it found bad ‘I saw Jean and Marc; the latter liked my idea, the former thought it was bad.’
She concludes, ‘-ci n’est pas très clair dans la langue’ (‘-ci isn’t very clear in the language’) (p. 143).
The idea that the demonstrative determiner and the locative reinforcer form a single constituent underlyingly is 8
particularly appealing in Spanish where, unlike what’s found in French, the two must be compatible in terms of a feature corresponding to proximity to the speaker (Brugè 2002: 27).
In (14) the functional structure between DefE and DemP is omitted for expository purposes. The important point is that 9
the noun raises over SpecDemP. For a slightly different movement analysis of the demonstrative determiners and their Thus, the term demonstrative is misleading. I retain it, though, since it’s the standard term for these determiners in the traditional and the generative syntactic literature.6
In order to capture the parallel between the non-deictic interpretation of the demonstrative determiner and the definite interpretation of the definite article, I propose to adopt a similar analysis: the non-deictic demonstrative determiner occupies DefE, the position to which the definite article raises from ClfE, if interpreted non-generically. This analysis is motivated below.
While the example in (12a) is ambiguous between a demonstrative and a definite reading, (12b) is unambiguously demonstrative:
(12) a. J’aime ce vin. b. J’aime ce vin-là. I-like this wine I-like this wine-there ‘I like this/the wine.’ ‘I like that wine.’
In (12b) the demonstrative determiner is reinforced with a postnominal deictic locative, distinguishing between distal/proximal, as in (13):7
(13) a. ces jours-ci b. ces jours-là these days-here these days-there
‘these days’ ‘those days’
In recent work on demonstratives, Brugè (2002) and Giusti (2002) propose that: (a) (true) demonstratives in all languages are phrasal constituents merged as the specifier of a functional projection (which I shall label demonstrative phrase, DemP) low down in DP* (but above NP*); (b) demonstratives raise from SpecDemP to the topmost specifier in DP* to check a referentiality feature against the highest head in DP*; (c) raising from SpecDemP to SpecDefP is overt or covert, depending on the strength of the referentiality feature: assuming that the head noun raises to a head higher than DemE, this means that, if demonstrative raising is covert, the demonstrative is postnominal, and if it’s overt, the demonstrative is prenominal.
There’s little in the syntax of the demonstrative determiners themselves in French to suggest such a movement-based approach; demonstrative determiners are always prenominal. However, this could simply be because the referentiality feature is strong in French and raising of the demonstrative determiner from
SpecDemP to SpecDefP is therefore overt. Furthermore, we find clear support for the Brugè/Giusti approach to demonstratives if we consider the syntax of the reinforcers -là/-ci, whose postnominal position is accounted for if they occupy SpecDemP. If, further, the surface position of -là/-ci, SpecDemP, is the underlying position of ce/cet/cette/ces, then, in a nominal like ce livre-ci ‘this book’, for example, the sequence ce-ci is merged as a constituent in SpecDemP. While 8 -ci stays put, ce raises to SpecDefP. Thus, unlike non-deictic demonstrative determiners, which are functional heads merged in DefE, their deictic counterparts (together with the locative reinforcer, if present) are merged (as single constituents) in SpecDemP, following Brugè’s (2002) proposal for Spanish. The surface order is derived, first, by raising the demonstrative determiner from its underlying position in SpecDemP to SpecDefP (leaving -ci/-là in situ), and, second, by raising the noun past -ci/-là:9
optional locative reinforcers see Bernstein (1997; 2001: 545–6).
The data are clearly more complex than suggested by Brugè (2002: 38) who considers the data in (i): 10
(i) a. ce livre-ci de Jean b. *ce livre de Jean-ci this book-here of J. this book of J.-here ‘this book of J.’s’
The examples in (15b, c) show that it’s not enough to say that de-marked nominal dependants remain in their thematic SpecèP positions within NP* (§2.1.3). (14) DefP ei Spec DefN g ri i ce DefE DemP g tu j
livre Spec DemN 4 tu i t -ci DemE NP* 6 j . . . t . . .
This analysis raises a number of questions regarding the linear placement of -ci/-là. It’s not just bare nouns that can intervene between the demonstrative determiner and the locative reinforcer in a structure like (14). Examples are given in (15):
(15) a. cette robe jaune-ci b. ce projet de loi-là this dress yellow-here this project of law-there ‘this yellow dress’ ‘that bill’
c. ces 3% de croissance-ci d. ce livre que tu m’as donné-là
these 3% of growth-here this book that you me-have given-there ‘this growth of 3%’ ‘that book you gave me’
The example in (15a) is unproblematic if attributive AP*s merge higher than DemE; the examples in (15b–d), in contrast, are very problematic. While we might argue that projet de loi in (15b) is a lexicalised complex noun which is merged under N and able to undergo N movement, such an approach isn’t without its problems, and, in any case, certainly can’t be adopted for 3% de croissance and livre que tu m’as donné in (15c, d). If -ci/-là are merged outside NP*, then what these latter examples suggest is that the movement which takes the noun and its dependants out of NP* is phrasal movement rather than head movement. Phrasal movement out of NP* into DP* is explored further in §3.7 in the context of attributive-AP* placement. If such movement is needed independently, then it makes sense to assume that that’s what underlies the word order in (15), too.10
In the above analysis, non-deictic demonstrative determiners (heads merged in DefE, the position to which truly definite articles raise) are distinguished from their truly deictic counterparts (phrases merged in SpecDemP which raise to SpecDefP). One might suppose, instead, that the demonstrative determiner is identical in both cases, and merged in DefE. The interpretative difference between the deictic and non-deictic interpretations would then hinge on -ci/-là. There is, however, hard semantic and syntactic evidence to support the dual analysis. First, ce can be interpreted deictically (albeit weakly), even without -ci/-là reinforcement. The second piece of evidence has to do with the extractability of a wh genitive. One of the parallels between DP* and CP* (§1.5) is that the highest specifier position is an escape hatch for wh fronting in both
(§§5.6.1ff.), possibly because of the notion of phase edge (Chomsky 2001): SpecCP* is an intermediate landing site for wh fronting out of a clause, as in (16a); SpecDP* is an intermediate landing site for wh fronting out of a nominal, as in (16b):
CP* i CP* i i
(16) a. [ Quand as-tu dit [ t que Jean est parti t ]]? when have-you said that J. is left
‘When did you say J. left?’
CP* i DP* i i
b. [ [De quels étudiants] as-tu corrigé [ t les copies t ]]? of which students have-you corrected the copies
‘Which students’ scripts did you mark?’
The example in (i) is fine: 11
(i) Dans quel livre as-tu lu ce chapitre-là? in which book have-you read this chapter-there ‘In which book did you read that chapter?’
Example (i) differs from text example (18a) in the preposition dans ‘in’. The difference crucially entails a different structural analysis, however. The PP dans quel livre ‘in which book’ is a VP* circumstantial of place rather than a thematic dependant of chapitre. As such, wh fronting doesn’t involve extraction from DP*. The presence of the demonstrative isn’t therefore expected to lead to ungrammaticality.
notes that they block wh extraction. Compare (17a) with (17b): (17) a. ¿De quién han publicado la foto?
of who have.3PL published the photo ‘Whose photo did they publish?’ b. *¿De quién han publicado esta foto?
of who have.3PL published this photo
The dual analysis of the demonstrative determiner in French – non-deictic merged in DefE, deictic raised to SpecDefP – allows us to make a prediction about the extractability of a wh phrase from a ‘demonstrative’ DP*: while wh fronting should be impossible with truly deictic demonstrative determiners (because the demonstra- tive determiner occupies SpecDefP, the all-important escape hatch), it should be fine with non-deictic
demonstrative determiners (since SpecDefP is free). The facts relating to deictic demonstrative determiners can be readily observed, since the deictic reading can be forced using a locative reinforcer. In contrast, those relating to the non-deictic reading are less accessible, since the absence of deixis isn’t formally marked. Nevertheless, the data suggest that the prediction is borne out. In (18a) the demonstrative determiner is clearly deictic (it’s reinforced with -là), and wh fronting is ungrammatical, as expected. In (18b), provided the demonstrative is non-deictic, wh fronting is possible, exactly as in (18c), where the definite article replaces the demonstrative determiner:11
(18) a. *De quel livre as-tu lu ce chapitre-là? (deictic reading) of which book have-you read this chapter-there
b. De quel livre as-tu lu ce chapitre? (non-deictic reading)
of which book have-you read this chapter ‘Which book did you read the chapter from?’ c. De quel livre as-tu lu le chapitre?
of which book have-you read this chapter (18b)
Thus, I conclude that while the deictic demonstrative determiner raises to SpecDefP, the non-deictic demonstrative determiner occupies DefE.