La LFTAIPG y la gestión de archivos en la Auditoría Superior de la Federación
2.2 La LFTAIPG y los archivos gubernamentales
Rizzi’s (1997) left periphery provides a template within which to see LD. TopicP is a ‘higher’ predication, above IP*: rather than being interpreted on the basis of è roles, checked by a regular lexical– predicative head, the topic and comment in SpecTopicP and CompTopicP in (92) are interpreted on the basis of the pragmatic roles they play, encoded as features checked against TopicE. In (83a) the LDed phrase is the subject Jean, resumed by the clitic il. Given the analysis of subject proforms in §4.4.4, Jean can’t occupy the canonical subject position because this position is occupied by the subject proform (prior to phonological cliticisation onto I*E). Instead, the LDed phrase occupies a specifier position outside IP*, namely, (Spec)TopicP with the comment in complement position:
(92) TopicP 3 topic TopicN Jean 3 TopicE comment il m’aime bien
A number of facts suggest that LD is the output of Merge rather than the result of movement of the topic out of the comment (pace De Cat 2002). First, the comment is always syntactically well formed even without the LDed topic. This suggests that the comment is a grammatically complete constituent, in a way that constituents containing an antecedentless trace are not. Second, the characteristic prosody of LD isn’t found in any clear case of movement. Third, the comment typically contains a resumptive clitic (provided one exists) correspond- ing to the LDed topic. If LD were the result of Move, this would mean there were two overt realisations of the same underlying constituent within IP* which would be surprising since the relevant varieties of French aren’t characterised by clitic doubling. Fourth, the dependency between the LDed phrase and any plausible
antecedent within the (relevant) comment isn’t subject to the constraints on movement found elsewhere: not only is it unbounded (unlike head movement), it can also cross an island (unlike wh movement). In (93a) the LDed topic is resumed by a clitic in the subordinate clause; in (93b) it’s resumed within a relative clause: (93) a. Moi, [il faut [que j’aille en ville]].
me it is-necessary that I-go to town ‘I need to go to town.’
A hanging topic is a discourse rather than sentence topic. 42
Ayres-Bennett and Carruthers (2001: 268, following Deulofeu 1977) call these ‘binary constructions’. They’re ‘very 43
informal’ according to Gadet (1997).
Note, though, that, unlike LDed argumental/stage topics, hanging (discourse) topics can’t iterate, must precede 44
argumental/stage topics, and are restricted to pre-matrix-clause position. Thus, they may well occupy a specifier position in the left clause periphery which is even higher than (Spec)TopicP.
b. Jean, [le jour [où il ne sera pas en retard] sera un jour de fête]! J. the day where he NEG will.be not in late will.be a day of celebration ‘The day J. isn’t late will be a day of celebration.’
Fifth, the nature of the correspondence between the LDed phrase and the resumptive proform in the comment is flexible. Thus, while in (83) and (84), the match is perfect – LDed subject resumed by subject proform, LDed object resumed by object proform – in (94) it isn’t:
(94) a. Jean, je lui ai donné €20. J. I to.him have given €20 ‘I gave J. €20.’
b. Tout le monde, le soir de la fête, on était tous fatigués. all the world the evening of the party one was all tired ‘On the evening of the party, everyone was tired.’
In (94a) the LDed phrase is resumed by an indirect-object clitic, but isn’t itself marked as an indirect object with à; in (94b) the LDed phrase is resumed by on ‘one’. Further cases of flexibility are illustrated in (95): (95) a. Ce métier | on se déplace tous les jours.
this job one self moves all the days ‘In this job you’re on the move every day.’ b. Le même argent | on peut payer un loyer.
the same money one can pay a rent ‘At that price you can afford to pay rent.’
The vertical line splits these sentences into two halves which are syntactically independent and only vaguely semantically related. There’s no resumptive clitic. The first half is called a hanging topic, the relationship42
between the two, ‘loose aboutness’. Intonation rises on the hanging topic, then falls on the rest of the43
sentence, as in LD, suggesting a parallel analysis. Given that it’s difficult to see what kind of strictly syntactic44
relationship the hanging topic might have with the rest of the clause, an extraction analysis is implausible, suggesting that we have a base-generated topic–comment articulation. Given the parallel with regular LD, a non-extraction analysis of LD is suggested, too. I conclude therefore that LD is a base-generated configuration rather than the result of movement. Zubizarreta (2001: 184) talks about LD involving its own predicate– argument structure: rather than being merged within the lexical VP* and subsequently raised, the dislocated phrase is, instead, merged outside the core clause as the ‘subject’ of the higher predication; from this position it binds an argument variable within the core clause.
To account for the fact that LD isn’t restricted to a single clause-initial topic, Rizzi (1997) claims that TopicP can iterate and that, in principle, there’s no limit to the number of topics that can be LDed. In (84a) there are two LDed topics and each is resumed within IP*. Consider, though, the multiple LD in (96):
(96) Moi, mon frère, sa femme, [elle est malade]. (after Yaguello 2003b: 220) me, my brother, his wife, she is ill
‘My brother’s wife is ill.’
The multiple LD in (96) differs from (84a) in that the last of the three LDed topics alone is resumed within the bracketed IP*. However, this difference is only significant if we assume that resumption is required within IP*. Note that there’s nothing in the topic–comment articulation in (92) which assigns any particular status to IP*; the comment is simply whatever occupies CompTopicP, and this doesn’t have to be IP*. Assuming iterability of TopicP, in cases of multiple LD like (96), the comment in all but the lowest topic–comment articulation is itself a further topic–comment articulation. Thus, what’s relevant is whether or not each successive topic is resumed within its particular comment. And, as shown in the representation of (96) in (97), this is indeed the case:
This observation is a further reason to reject the extraction analysis of LD. 45 (97) Topic1P ei topic Topic1N g 3 i
Moi Topic1E comment g Topic2P wo topic Topic2N g 3 i j
[mon frère] Topic2E comment g Topic3P wo topic Topic3N g 3 j k
[sa femme] Topic3E comment g IP* 6
k
elle est malade
In the highest topic–comment articulation, Topic1P, the topic moi ‘me’ is resumed by the possessive
determiner within mon frère ‘my brother’, contained within the relevant comment, Topic2P, the complement of Topic1E; in Topic2P, mon frère is resumed by the possessive determiner within sa femme ‘his wife’, contained within Topic3P; finally, sa femme is resumed by elle ‘she’, the subject of the core clause. Thus, just as in (84a), each LDed topic in (96) is resumed within the relevant comment (even if this isn’t necessarily IP*).
The order in which topics are resumed within the comment has no impact on the order in which they are LDed (contra Larsson 1979). However, the inherently hierarchical model of LD in (97) means that the linear45
order of LDed phrases does determine their relative topichood. While moi, mon frère and sa femme in (96)/ (97) are all topics within their own topic–comment articulation, they don’t all have the same pragmatic status relative to one another; there’s a left-to-right pecking order, which reflects the structural hierarchy of shifting topics: within the context of the entire utterance moi is more topical than sa femme. Such an approach suggests that the multiple LD in (84a) is similarly hierarchical: that is, rather than being two randomly ordered topics of equal standing, Jean is more topical than Marie, as a simple consequence of the correspondence between linear order (of LDed topics) and hierarchical pragmatic structure. Similarly, while (98a, b) are truth- conditionally equivalent, (98a) doesn’t correspond pragmatically to (98b):
(98) a. Jean, Marie, il la voit demain. b. Marie, Jean, il la voit demain. J. M. he her sees tomorrow M. J. he her sees tomorrow ‘J.’s seeing M. tomorrow.’ (98a)
The sentences in (98a, b) are formally identical apart from the relative order of the LDed phrases: both are topics, but the topichood of the leftmost LDed phrase is greater than that of the rightmost. Thus, topichood is a relative or scalar notion, rather than a categorical one.
Topic1P i Topic2P j IP* i j
(99) a. [ [Jean] Topic1E [ [Marie] Topic2E [ il la voit demain]]] (98a)
Topic1P j Topic2P i IP* i j
b. [ [Marie] Topic1E [ [Jean] Topic2E [ il la voit demain]]] (98b)
5.3.2
Right dislocation
Dislocation is also possible to the right: (100) a. Jean lui a donné les €20, à Marie.
b. Jean les a donnés à Marie, les €20. c. Il a donné les €20 à Marie, Jean.
Despite Harris’s (1978: 119) claim that RD is grammaticalising (and therefore increasingly pragmatically neutral), 46
Ashby (1982) finds no speaker–variable evidence to indicate a change in progress.
Cecchetto’s analysis of clitics differs from the one set out in §4.4.1. However, the differences are irrelevant for our 47
purposes here.
As with LD, argumental RDed topics are usually resumed with an appropriate clitic, as in (101a), unless it’s generic, in which case resumption is optional, as in (101b):
(101) a. Tu les aimes bien, les colliers? b. Tu aimes bien (ça), les colliers? you them like well, the necklaces you like well that, the necklaces ‘Do you like the necklaces?’ (specific) ‘Do you like necklaces?’ (generic)
RD differs from LD in phonological and syntactic as well as pragmatic terms. In phonological terms, RD doesn’t show the characteristic ‘comma intonation’ of LD. The RDed phrase isn’t prosodically prominent, and is actually destressed. Syntactically, an RDed phrase corresponding to an indirect object or an inherent-case- marked DP* typically usually bears overt inherent case marking, as in (102) (cf. (94a)):
(102) a. On va les avoir, les Allemands, on va leur montrer, aux Anglais! one goes them have the Germans one goes to.them show to.the English ‘We’ll beat the Germans, we’ll show the English!’
b. J’en ai acheté, du lait. I-of.it have bought of.the milk ‘I bought some milk.’
More generally, RD is more restricted than LD. According to Lambrecht RD is more presuppositional than LD, LDed constituents are never contrastive/emphatic, RD is compatible with known topics, only. Thus, ‘loose aboutness’, sufficient for an LDed hanging topic, as in (95), isn’t enough in RD. Certainly, the pragmatic46
function of RD is less clearly delineated than that of LD.
Unlike what we saw with LD (§5.3.1), an analysis of RD can’t straightforwardly exploit Rizzi’s (1997) CP* for the simple reason that CP* is left peripheral while RD ‘happens’ on the right. The problem is addressed by Cecchetto (1999) (using data primarily from Italian). Cecchetto rejects (p. 41) the idea (now admittedly largely abandoned, but see De Cat 2002) that LD and RD are configurationally identical, differing only in the
direction in which SpecTopicP branches, since such an approach fails to account for a number of empirical differences between RD and LD which aren’t predicated to exist assuming configurational identity. He also rejects (pp. 48–55) an (at that time unpublished) reanalysis of RD of Richard Kayne’s in terms of repeated LD. On such an approach, in addition to existing in its own right, LD also feeds the derivation of RD via a second instance of LD fronting the remnant IP*.
Instead, Cecchetto (1999) relates RD to LD and the clitic-doubling phenomenon found in several Romance varieties. He suggests (pp. 56–62) that, in the initial stages of the derivation, the three constructions are identical: they all involve raising of an XP from its base position to the specifier of a left-VP*-peripheral projection whose head hosts the clitic. At that point the three constructions diverge: the clitic double moves47
no further, resulting in regular IP*-internal clitic doubling; the LDed constituent raises to Rizzi’s (1997) CP*- internal SpecTopicP position; and, the RDed constituent raises to a left-VP*-peripheral SpecTopicP position (Belletti 2004a).
Ceccheto’s (1999) analysis of RD is difficult to reconcile with: (a) the non-movement approach to LD (§5.3.1); and (b) the analysis of non-subject clitics in §4.4.1. What’s clear is that RD isn’t random right adjunction. Not only would such an approach sit awkwardly alongside the kinds of structures seen elsewhere, it would fail to account for the hierarchical properties of RD (parallelling those of LD; cf. (97)) illustrated in (103) (from De Cat 2002: 16, (1b)):
(103) Elle était vraiment bien, son exposition, à Julia. it was really good, her exhibition, to J.
‘J.’s exhibition was really good.’
5.4
Focus fronting
Focus fronting in English is illustrated in (104): (104) My dad I called yesterday.
Rizzi’s (1997) CP* provides a structural template within which to analyse focus fronting, namely, a focus– presupposition articulation, structurally encoded as a FocusP, whereby the focus raises to SpecFocusP and the presupposition occupies CompFocusP:
(105) FocusP ei focus FocusN My dad ei FocusE presupposition I called yesterday
In contrast to English, French doesn’t allow focus fronting: (106) *Mon papa j’ai appelé hier.
my dad I-have called yesterday
Zubizarreta (2001: 184) accounts for the ungrammaticality of examples like (106) by claiming that, unlike English, French simply lacks FocusP. However, the issue is somewhat more complex, and a number of factors are involved. The first is stylistic. Clear-cut examples of focus fronting, like (107), are grammatical and characteristic of high register ModF rather than low register ConF:
(107) A Jean j’ai donné €20. to J. I-have given €20 ‘I gave J. €20.’
Thus, Zubizarreta’s conclusion about the absence of FocusP may apply specifically to ConF rather than to ModF. A movement approach to the example in (107) is suggested by a number of factors, all of which contrast with LD: the left-peripheral phrase is necessarily case-marked; there’s no ‘comma’ intonation; no resumptive clitic appears within IP*; the fronted phrase is focal rather than topical. Further, De Cat (2002: 81) notes that, again unlike LD, focus fronting is sensitive to islands and weak crossover effects and existential indefinites can undergo focus fronting.
While focus fronting is characteristic of ModF, something resembling focus fronting is found in ConF, and is more common that is sometimes acknowledged. In fact, there are two separate phenomena to account for. First, there are cases of informational focus, as in (108), from Blanche-Benveniste (2003: 342):
(108) a. 10F, vous auriez? b. Même de moi, il se méfie. 10F you would.have even of me he self mistrusts ‘Would you happen to have 10F?’ ‘He’s wary even of me.’ c. Trente ans, elle avait. d. Pour un roi, il se prenait.
thirty years she had for a king he self took
‘She was thirty.’ ‘He behaved like he was royalty.’
The ‘comma’ intonation found here suggests we’re dealing with LD, albeit without resumptive proforms. Kroch (2001: 712) discusses the example in (109):
(109) 10F, ce truc m’a coûté. 10F this thing me-has cost ‘This thing cost me 10F.’
He claims that such structures had a single intonation contour in OF, but ‘comma intonation’ since MidF. He concludes that these are now instances of LD. Goosse (2000: 134) contrasts (110a) with (110b):
(110) a. Mon erreur, je la reconnais. b. Mon erreur, je reconnais. my mistake I it recognise my mistake I recognise ‘I recognise my mistake.’ = (110a)
Example (110a) is a regular case of LD, while in (110b) there’s no resumptive clitic. Examples like (110b) are often considered characteristic of Belgium or Flanders. However, Goosse suggests that such structures exist in France, too, and indeed are becoming increasingly common, suggesting that focus fronting à l’anglaise may be gaining ground.
Or indeed eux/elles ‘them M/F’, but not 1PL nous or 2PL vous: 48
DP* DP*
(i) a. Ce sont [ eux/elles] qui rient. b. *Ce sont [ nous/vous] qui rions/riez. it are them.M/F that laugh it are us/you that laugh
‘They are laughing.’
(111) a. Les haricots, j’aime pas. b. Les poireaux, je déteste. the beans I-like not the leeks I detest Beans, I don’t like. ‘Leeks, I hate.’
Again, the intonation patterns point to LD without resumption. Rizzi (1997) notes that LD without resumptive proforms is possible with a restricted class of verb (as well as some prepositions), only, which can select a non- overt referential proform. (See also Zribi-Hertz 1984.) Interestingly, the members of the ‘restricted class of verb’ which allow structures like those in (111) are the same as those discussed in §5.2.3 which select a bare infinitive which can be ellipsed (rather than pronominalised), namely, adorer ‘to adore’, aimer ‘to like’, désirer ‘to desire’, détester ‘to detest’, préférer ‘to prefer’ and vouloir ‘to want to’. It makes sense, then, to attribute the grammaticality of the examples in (111) to the same abstract feature with allows these verbs to ellipse their bare-infinitival dependant.
In summary, focus fronting is the result of movement and is found in ModF but not ConF. Superficially similar constructions which are found in ConF are in fact base-generated structures akin to LD.
5.5
Clefting and pseudo-clefting
While true focus fronting is restricted to ModF, and unavailable in ConF (§5.4), there’s one focalisation device available in both varieties, namely, (pseudo)clefting, illustrated in (81d, e) on page 123. These structures have been attested since the OF period and have always had marked pragmatics. The examples in (112) illustrate clefts:
(112) a. C’est Luc qui veut te voir. b. Ce sont mes parents que je vois demain. it is L. that want you see it are my parents that I see tomorrow ‘L. wants to see you.’ ‘I’m seeing my parents tomorrow.’
Clefts take the form Ce [être] [XP] que/qui [IP*], whereby XP is termed the focus, IP*, the coda. A number of comments are in order:
– Only simple forms of être are possible (Table 2.1 on page 21 in §2.2.1.2). The most common are present, imperfect and future, as in (113a–c). Compound verb forms (Table 2.3 on page 22) are excluded, as shown in (113aN–gN):
(113) a. C’est L. que/qui . . . aN. *Ç’a été L. que/qui . . . b. C’était L. que/qui . . . bN. *Ç’avait été L. que/qui . . . c. Ce sera L. que/qui . . . cN. *Ç’aura été L. que/qui . . . d. Ce serait L. que/qui . . . dN. *Ç’aurait été L. que/qui . . . e. Ce fut L. que/qui . . . eN. *C’eut été L. que/qui . . .
f. (Il faut que) ce soit L. que/qui . . . fN. *(Il faut que) ç’ait été L. que/qui . . . g. (Il fallait que) ce fût L. que/qui . . . gN. *(Il fallait que) c’eût été L. que/qui . . . – In ConF SGce can be replaced with ça where the form of être prevents elision: (114) a. %Ça serait trop bien que tu reviennes. b. %Ça sera lui qui partira.
that would.be too good that you return that will.be him who will.leave ‘It’d be great if you came back.’ ‘He’s the one who’ll leave’.
– In ModF, where the focus is a PL lexical DP*, 48être is PL, as in (115a–g); in ConF être remains SG, as in (115aN–gN):
ModF ConF
(115) a. Ce sont Luc et Anne que/qui . . . aN. C’est . . . b. C’étaient Luc et Anne que/qui . . . bN. C’était . . . c. Ce seront Luc et Anne que/qui . . . cN. Ce sera . . . d. Ce seraient Luc et Anne que/qui . . . dN. Ce serait . . . e. Ce furent Luc et Anne que/qui . . . eN. Ce fut . . .