The evolution of negation in French and Italian

0 downloads 0 Views 197KB Size Report
Nov 1, 2012 - blush, the explanation might seem very simple: as the above quote shows, and as remarked ...... shame or offence in the court my. 'And I would ...
The evolution of negation in French and Italian: Similarities and differences1 Maj-Britt Mosegaard Hansen & Jacqueline Visconti University of Manchester/University of Genoa

This article examines similarities and differences in the evolution of both standard clause negation and n-word negation in French and Italian. The two languages differ saliently in the extent to which standard negation features postverbal markers. We suggest that a convergence of phonetic, prosodic, morphosyntactic, and pragmatic changes in the evolution of French may explain why the grammaticalization of the postverbal marker is significantly more advanced in that language. Two types of n-word negation must be considered: (i) those where the n-word occurs postverbally, and (ii) those where an n-word is positioned preverbally. In the former type, French allows deletion of the preverbal marker, whereas Italian does so to a much lesser extent. In the second type, French allows (indeed, normatively demands) insertion of a second preverbal negative marker, whereas Italian does not. We suggest that this is attributable to the respective positive vs negative etymologies of the n-words. In type (i) constructions, this etymological difference appears to make Italian a negative-concord language from the outset. In contrast, negative concord in Modern French has, to a large extent, developed gradually out of what was originally a reinforcement of standard negation by positive items with scalar properties. Our analysis suggests that the pace and form of grammaticalization cannot be attributed to any single cause, but is rather the result of a confluence of formal and functional factors. Keywords: Negation, Jespersen’s cycle, standard negation, n-words, grammaticalization, Romance The authors would like to thank the special issue editors, Anne Carlier, Walter de Mulder, and Béatrice Lamiroy, for inviting us to contribute to the issue, and for all the work they have done in bringing the project to fruition. We are also very grateful to the two anonymous referees whose comments helped improve the article. The usual disclaimers apply. 1 

Folia Linguistica 46/2 (2012), 453–482.   doi 10.1515/flin.2012.016 issn 0165–4004, e-issn 1614–7308  © Mouton de Gruyter – Societas Linguistica Europaea Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

454   Maj-Britt Mosegaard Hansen & Jacqueline Visconti

1. Introduction The aim of this article is to examine the evolution of negative strategies from Latin to present-day French and Italian, with reference to cyclical developments of the type first noticed by Jespersen (1917). Our focus will be on the most basic forms of negation used in declarative main clauses, viz., firstly, what Payne (1985) calls standard negation, that is, the most basic form of clause negation, which applies productively in declarative main clauses and does not involve quantifiers (e.g. Fr (ne) . . . pas, It non ‘not’), and secondly, negative clause constructions involving indefinite quantification by so-called n-words (e.g. Fr (ne) . . . personne, It non . . . nessuno ‘nobody’).

2.  Standard negation in French and Italian Starting with standard negation, it is a salient feature of this type of negation that its formal expression tends to evolve and change over time. Of particular interest in the context of grammaticalization is the fact that, cross-linguistically, the relevant changes appear very frequently to follow cyclical patterns. The most famous such pattern identified in the literature is the so-called Jespersen’s Cycle (so dubbed by Dahl 1979: 88, based on Jespersen 1917). Jespersen’s Cycle pertains to a number of languages which express standard negation through one or more markers of an adverbial nature. Jespersen himself describes the cycle as follows: [t]he original negative adverb is first weakened, then found insufficient and therefore strengthened, generally through some additional word, and this in turn may be felt as a negative proper and may then in the course of time be subject to the same development as the original word.  (Jespersen 1917: 4)

The evolution of French, in an extended sense including its mother language Latin as well as a number of French-based Creoles, nicely illustrates the full sequence, as set out in Table 1. With respect to standard negation, French and Italian differ saliently with respect to their current stage(s) in the cycle, that is, in the extent to which postverbal negative markers are used at all, let alone conventional-

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

The evolution of negation in French and Italian    455 Table 1.  The evolution of French clause negation (sample sentence: ‘I do not say . . .’) Stage 0. [Classical Latin] Stage 1.

non dico je ne dis

Stage 2.

je ne dis (pas/ mie/point)

Stage 3.

je ne dis pas

Stage 4.

je (ne) dis pas

Stage 5. [Future French?] je dis pas Stage 6. [Louisiana French mo pa di Creole]a

The negator is preverbal The preverbal negator is phonetically reduced The preverbal negator is optionally complemented by a postverbal element The postverbal element grammaticalizes as part of a discontinuous negator embracing the verb The original preverbal negator becomes optional The negator is postverbal The previously postverbal negator migrates to preverbal position

  As pointed out by an anonymous referee, we cannot of course assume that Creoles necessarily represent future stages of the superstrate language. Louisiana French Creole is thus adduced principally to illustrate the full Jespersen Cycle. That said, French negated infinitives have in fact already completed the cycle, as shown in (i) (see also Martineau 1994): a

(i) Il m’ a dit de (ne) pas y aller. SBJ.CL-3SG DAT.CL-1SG have-PRS-3SG said-PST.PTCP to (NEG NEG there-CL go-INF ‘He told me not to go there.’

ized. Thus, standard, particularly written, French has been consistently at Stage 3 since the seventeenth century, while colloquial, particularly spoken, French is currently very firmly at Stage 4, some dialects (particularly Québecois, see Sankoff & Vincent 1977) showing strong signs of progression to Stage 5. Standard Italian, on the other hand, has essentially remained at Stage 0, although in the colloquial register – and particularly in northern varieties2 – it allows Stage 2 constructions, optionally adding a postverbal More precisely (but see e.g. Parry, in press, for a more detailed picture): preverbal negation, continuing Latin morphology and syntax, is found in standard Italian (i), in the Florentine dialect on which it is based, in the other central and southern dialects of Italy and Sardinia, as well as in Venetan, Friulian and Ligurian (ii). Bipartite negation is found in Emilian (iii), as well as some alpine Lombard (Ticinese) and border Piedmontese-Ligurian varieties, while Piedmontese and Lombard have mainly postverbal negation (iv); the postverbal element may originate in a noun with general reference merged with a negative, such as Piedmontese nen(t) < NEC ENTE(M) ‘no being’ or < NE GENTE(M) ‘no people’ 2 

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

456   Maj-Britt Mosegaard Hansen & Jacqueline Visconti

mica (see (1)). Note, however, that this development has taken place without the intervening Stage 1 that we find in French: (1) Non dico mica . . . NEG say-PRS-1SG NEG ‘I do not say [mica] . . .’

Consequently, the syntactic status of the preverbal marker is not the same in the two languages. While the preverbal non is very clearly the principal negator in Italian, even in examples like (1), Rowlett (1998: ch. 1) argues cogently that in Modern French (i.e. from Stage 3 onwards) that role is played by the postverbal pas, the preverbal ne having been reduced to a type of agreement marker. As such, ne is insufficient to mark negation on its own in all but a very restricted set of syntactic contexts in elevated registers, and it has purely expletive uses in certain other contexts, as in (2). Pas, on the other hand, has inherent negative meaning in all contexts and registers, independently of the presence of ne. (2) J’ ai peur qu’ il SBJ.CL-1SG have-PRS-1SG fear that SBJ.CL-3SG ne le fasse. NEG.EXPL OBJ.CL do-PRS.SBJV-3SG ‘I fear that he may (*not) do it.’

We are thus faced with a puzzle: given the close genetic and areal relations between French and Italian, why has the expression of negation evolved so much more rapidly in the former than in the latter? At first blush, the explanation might seem very simple: as the above quote shows, and as remarked upon by van der Auwera (2010: 80), Jespersen (1917) (Rohlfs 1968: 218), in a minimiser such as Lombard mi(n)ga (< MICA(M) ‘crumb’), or, less frequently, in a resumptive holophrastic negator added clause-finally, such as Lombard nò (see e.g. Bernini & Ramat 1998; Ramat 2006b: 360f): (i) Non dormirò. (Italian) NEG sleep-FUT-1SG (ii) Nu durmiò. (Ligurian) NEG sleep-FUT-1SG (iii) A n dorum briza. (Emilian (Modena)) NEG sleep-FUT-1SG NEG (iv) Dürmirai nen. (Piedmontese) sleep-FUT-1SG NEG ‘I shan’t sleep’ (AIS, Map 653)

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

The evolution of negation in French and Italian    457

essentially explains the negative cycle as being triggered by the phonetic weakening of the original preverbal marker. As no such phonetic weakening has taken place in Italian, unlike in French, there would seem to be no a priori reason for that language to progress beyond Stage 0 of the cycle. In our view, an explanation along these lines would be too simplistic, however, for the following reasons: First of all, as shown by (1), Italian does offer the option of adding a postverbal marker despite the fact that the preverbal marker has remained phonetically substantial, so Stage 1 is clearly not a pre-requisite for a language to reach Stage 2. Indeed, according to Kiparsky & Condoravdi (2006: 4), the evidence for phonetic weakening of preverbal negators in languages that participate in Jespersen’s Cycle is in fact not very strong cross-linguistically. Even French is not a thoroughly convincing case, in as much as ne only lost its ability to carry stress towards the end of the Middle French period (16th c.), i.e. approximately half a millennium after the language entered Stage 2 (cf. Martineau & Mougeon 2003: 123f). For Italian too, although the varieties which have progressed beyond Stage 2 are the Gallo-Italian varieties, which are subject to vocalic weakening (NON > ne, n) (possibly due to a Celtic substratum), such weakening is characteristic of several southern varieties, too, for example, Abruzzan and Campanian (see Vai 1995: 164–5 for arguments against the hypothesis that phonetic erosion contributed to the loss of the preverbal negative in northern Italy). Moreover, Biberauer (2009) has shown that in at least one language – contemporary Afrikaans – the innovative postverbal negative marker is phonetically weaker than the original preverbal negator. So, if the formal properties of negation do not suffice to explain the cycle, what other parameters might be relevant? Observing that Latin NŌN itself is assumed to represent the univerbation of IE *ne + OENUM (‘one’), Jespersen’s contemporary Antoine Meillet proposed instead (prior to the publication of Jespersen’s work) that pragmatics might be driving the evolution of standard negation: Les langues suivent ainsi une sorte de développement en spirale: elles ajoutent des mots accessoires pour obtenir une expression intense; ces mots s’affaiblissent, se dégradent et tombent au niveau de simples outils grammaticaux; on ajoute de nouveaux mots ou des mots différents en vue de l’expression; l’affaiblissement recommence, et ainsi sans fin. (Meillet 1921[1912]: 140) ‘Languages thus undergo a sort of spiral development: they add extra words to obtain an intensified expression; those words weaken, wear out and are

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

458   Maj-Britt Mosegaard Hansen & Jacqueline Visconti reduced to the level of simple grammatical tools; new or different words are added for expressive purposes; the weakening process begins anew, and so on without end.’ [Our translation.]

There are a number of reasons to think that Meillet’s account may be more accurate, and certainly, from the point of view of grammaticalization theory, we would expect the formal changes that characterize Jespersen’s Cycle to be driven by the meanings attached to different constructions rather than the other way around. Thus, neither the transition from simple preverbal negation to bipartite, embracing negation nor the transition from bipartite to simple postverbal negation is achieved abruptly: as Table 1 suggests, both are characterized by periods of variation, represented by Stages 2 and 4. Given that, in some cases, these stages last for centuries and appear quite stable, it is plausible to assume that the competing strategies may, at least initially, differ not just in form but also in aspects of their meaning. Indeed, contemporary descriptions of Italian Stage 2 varieties consistently treat the bipartite construction non . . . mica as pragmatically different from the plain preverbal negator (e.g. Cinque 1976, Molinelli 1988, Bernini and Ramat 1996, Parry 1996, in press; Zanuttini 1997), as is also the case with other Romance vernaculars that are currently at Stage 2 (e.g. Catalan no . . . (pas), see Espinal 1993, and Brazilian Portuguese naõ . . . (naõ), see Schwenter 2006). At an intuitive level, the pragmatic difference in question seems to have to do with somehow emphasizing negation. Now, far from being contingent on formal weakening of the canonical negator, the ability to express negative emphasis would appear to be a universal feature of languages (Schwegler 1988: 36; Kiparsky & Condoravdi 2006: 7), and it is moreover a category that is commonly formally marked by the addition of a particlelike item. Frequent etymological sources for reinforcing postverbal markers have in common that they are elements which naturally lend themselves to the expression of emphasis, for instance quantitative expressions denoting minimal amounts of something, i.e. expressions that either are negative polarity items or have NPI-like properties. Saliently, this is the case with both French (and Catalan) pas (< Latin PASSU(M) ‘step’), and with Medieval French mie and its Italian cognate mica (< MICA(M) ‘crumb’). As (3) and (4) show, this type of expression can be found reinforcing negation from Pre-Classical Latin onwards, in texts having an oral and/or colloquial tenor:

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

The evolution of negation in French and Italian    459 (3) AM. Haec sacerdos Veneris hinc me petere aquam iussit a vobis. SC. At ego basilicus sum: quem nisi oras, guttam non feres. who-ACC if-not entreat-PRS-2SG, drop-ACC NEG carry-FUT-2SG ‘AM. That priestess of Venus ordered me here to get some water from you. SC. But I’m an important person: if you don’t entreat me, you’ll not take a drop away.’ (Plautus, Rudens, 3rd–2nd c. BC) (4) Quinque dies aquam in os suum five days-ACC water-ACC in mouth-ACC his-ACC non coniecit, non micam panis NEG put-PFV.PST-3SG NEG crumb-ACC bread-GEN ‘for five days he didn’t put any water in his mouth, not a crumb of bread.’ (Petronius, Satyricon, 1st c. AD)

Given examples such as these, it is plausible to assume a three-stage development from Latin to Medieval French/Modern Italian: at the initial stage, we have a variety of ordinary lexical items with no inherent preference for either positive or negative contexts, but which – because they happen to denote minimal quantities – lend themselves quite naturally to underscoring the negative content of a clause with a semantically compatible verb, following a rhetorical strategy that Detges & Waltereit formulate as follows: If you want to express in a strong way that some state of affairs did not take place at all, say that the state of affairs in question did not even take place to the smallest degree imaginable.  (Detges & Waltereit 2002: 177)

By the medieval period already, use of pas/mie and mica is, however, no longer constrained to certain types of verbs, and these items have (largely) lost their nominal properties in favor of a particle-like status, occurring without determiners or any other form of modification; see (5) and (6): (5) Soleill n’ i luist ne blet sun NEG there shine-PRS-3SG NEG corn n’ i poet pas creistre, NEG there can-PRS-3SG NEG grow-INF ‘The sun doesn’t shine, and corn cannot [pas] grow,’ (Chanson de Roland, 11th c., v. 980)

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

460   Maj-Britt Mosegaard Hansen & Jacqueline Visconti 3 (6) La grandezza delle magioni non cessa mica la febbre  the greatness of:the houses NEG stop-PRS-3SG NEG the fever ‘The greatness of a home does not [mica] stop the fever’. (Tesoro volg., ed. Gaiter, XIII ex. (fior.), 8–14 [TLIO]).

While occasionally found alongside NPIs, which strongly suggests that speakers must (have) perceive(d) them as part of the negation itself (cf. (7)(8)), these particles also sometimes occur without a preverbal negator and with non-negative meaning (cf. (9)-(10)): (7) Tuit vos Franceis ne valent pas meaille. all your-PL Frenchmen NEG be-worth-PRS-3PL NEG small-coin ‘All your Frenchmen aren’t [pas] worth a dime.’ (Li coronemenz Looïs, 12th c., v. 2433) (8) Di cinnamomi, cardamomi, curcume, zenzeri, garofani e cannelle of cinnamon trees cardamom curcuma ginger cloves and cinnamon non so mica un bel niente NEG know-PRS-1SG NEG a nice nothing ‘About cinnamon, cardamom, curcuma, ginger I don’t know [mica] a thing’. (http://freeforumzone.leonardo.it/lofi/curcuma/D2232404.html) (9) S’ il daigne pas parler ancor. if he-NOM deign-PRS-3SG NPI speak-INF still ‘If he still deigns to speak at all [pas].’ (Le roman de Renart, 12th–13th c., 12636–7; cited in Price 1993 : 193) (10) e se ‘l prenze dell’oste di questo cotale, che dett’è, dottasse né mica ch’ elli doubt- IPFV.PST.SUBJ-3SG NPI that he non fusse leale, NEG(-EXPL) be-IPFV.PST.SUBJ-3SG loyal sì gli die méttare guardie indosso (Egidio Romano volg. 1288 (sen.) 3.3.11 [TLIO]). In the case of both French and Italian, the first stages of the grammaticalization of the minimizers, that is, their extension beyond ‘harmonious’ semantic contexts, such as those in (3) and (4) (bleaching) predate the first texts (Hansen & Visconti 2009: 147f, Parry, in press). Thus, in the oldest extant texts in both languages, we find pas/mie/mica co-occurring with direct objects, as in (6), although occasional use in partitive constructions such as (i) reveal their nominal origins: (i) Lá no se sente miga de male. there NEG REFL feel-PRS-3SG crumb of ill ‘There no ill at all is felt’. (13th c. Lombard, Pietro da Bescapè, p.72: 2435)

3 

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

The evolution of negation in French and Italian    461 ‘And if the prince of the army of this said man doubted in any way [mica] that he were (not) loyal, so he must put guards about him.’

Contexts in which the latter occurs are usually those classified at negative-polarity contexts (Ramat 1994: 2771; Marchello-Nizia 1997: 306; Eckardt 2003: ch. 4), pointing to the existence of an intermediate stage at which the postverbal particles must have grammaticalized into NPIs, before developing into negative particles proper. As is well-know, this kind of divergence, or layering, in grammaticalization (Hopper 1991: 24), where older uses of an item or construction remain alongside a newer function, is far from unusual. In French, the NPI uses of pas/mie/point, however, eventually disappeared during the Classical period (17th–18th c.), plausibly due to the above-mentioned change in syntactic status of the postverbal negator, from a pragmatically conditioned optional element to being the principal negator. For Italian mica, such uses are no longer found after the sixteenth century. If it is, in fact, increasingly frequent use of ‘emphatic’ postverbal marker(s), rather than weakening of the preverbal marker, that is the driving force behind Jespersen’s Cycle, the next question is whether we can be more specific about the exact nature of this form of emphasis at the stage where one or more postverbal elements have become grammaticalized as negative particles, but are still in free variation with simple preverbal negation. For, as Schwenter (2006) points out, emphasis is as such a rather illdefined notion. Several studies have suggested that the use of bipartite Stage 2 negation in various Romance languages is subject to discourse-functional constraints linked to the presupposed or otherwise given nature of the negated proposition or its underlying positive counterpart (Cinque 1976, Bernini & Ramat 1996, Espinal 1993, Schwegler 1988, Zanuttini 1997, Schwenter 2006, Visconti 2007, Hansen & Visconti 2009, Hansen 2009). Thus, according to Zanuttini (1997: 61), for instance, only (12) below would constitute a felicitous exchange with the addition of mica, whereas (11) would not, assuming that Gianni is not a contextually salient potential chauffeur: (11) A. Chi viene a prenderti? B. Non so. Ma Gianni non ha (*mica) la macchina. but Gianni NEG have-PRS-3SG (NEG the car ‘A. Who’s coming to pick you up? B. I don’t know. But Gianni doesn’t (*mica) have the car.

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

462   Maj-Britt Mosegaard Hansen & Jacqueline Visconti (12) A. Chi viene a prenderti? – Gianni? B. Non so. Ma Gianni non ha mica la macchina. ‘A. Who’s coming to pick you up? – Gianni? B. I don’t know. But Gianni doesn’t [mica] have the car.

Schwenter (2006) suggests that Brazilian Portuguese naõ . . . naõ only appears in propositions that are already activated in the short-term memory of the hearer, while Hansen & Visconti (2009) and Hansen (2009) propose that Medieval French clauses negated by ne . . . pas/mie were constrained to be discourse-old in the sense of Birner (2006), i.e. either already activated in the short-term memory of the hearer or inferable (i.e. accessible to activation) based on other propositions thus activated. In all the relevant Romance languages, the plain preverbal markers, on the other hand, are compatible with propositions conveying information that is entirely new to the discourse. If French, unlike Italian, Catalan, and Brazilian Portuguese, has proceeded to Stage 3 of Jespersen’s Cycle, this may in part be attributable to the fact that the bipartite negator in Medieval French could mark propositions that were merely inferable. A  pragmatic constraint including inferable propositions may be more easily loosened to embrace pragmatically neutral negation than one which includes only directly activated propositions. As noted by Hansen (2009: 242), inferable propositions are the numerically most important category negated by ne . . . mie/pas, and it is conceivable that not all hearers may have gone to the trouble of performing the necessary inferences on every occasion. Furthermore, Hansen & Visconti (2009) and Hansen (2009) show that the negative reinforcers were frequently used in Medieval French in contexts that were, so to speak, Janusfaced, in the sense that, while the negated proposition was discourse-old and hence backwards oriented (i.e. towards prior discourse), it could at the same time be interpreted as forwards oriented by expressing a contrast with the immediately following clause. This is exemplified by (13), where the negated clause constitutes a direct rejection of the previous speaker’s request, but simultaneously contrasts with the prediction made in the following clause: (13) « Biau Sire, fet Gauvains, donc me poez vos bien dire, s’il vos plest, en quoi sui tiex come vos me metez sus. » – « Je nel vos dirai mie,   I-NOM NEG:it-ACC you-DAT say-FUT-1SG NEG

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

The evolution of negation in French and Italian    463 fet cie, mes vos troveroiz par tenz qui vos le dira. » ‘“Good Sir, says Gawain, then you can surely tell me, if you please, in what way I am that which you accuse me of.” – “I’ll not [mie] tell you, he says, but soon you’ll find one who will tell you.” ’ (La queste del Saint Graal, 13th c., p. 52)

Such instances leave hearers free to reinterpret the bipartite negative expression as an unmarked form of negation, not subject to any activation constraints. A couple of additional factors may have contributed to the obligatorification of postverbal negators in French, as opposed to Italian. One plausible factor, which is formal in nature, but which has significant pragmatic impact, is a prosodic change that took place during the medieval period. French, like Latin and other Romance languages, originally had wordbased primary stress, such that any word in a sentence or clause could in principle receive prosodic emphasis. By the sixteenth century, that was no longer the case; instead, primary stress had become tone-group based, falling by default on the last syllable of a tone group (Picoche & MarchelloNizia 1989: 184). This change, which appears to have taken several centuries to complete, would have made it increasingly difficult to stress the preverbal ne, and as noted above, the latter indeed lost its ability to carry stress towards the end of Middle French, its vowel being reduced to a schwa. The postverbal markers, on the other hand, would or could quite naturally receive primary stress in many, perhaps the majority of cases (intransitive clauses, transitive clauses with preverbal pronominal objects or compound tenses). Clause negation being a semantically and pragmatically very important function, it would not be surprising if the use of bipartite negation increased in consequence. A final, more speculative, factor that may have contributed to the transition from Stage 2 to Stage 3 is language contact. Thus, Medieval French was in close contact with several Germanic languages (English, German, and Dutch) all of which completed the transition from Stage 1 to Stage 5 of Jespersen’s Cycle in the course of the medieval/Renaissance period (cf. discussion in Ramat 1984: 178ff).4 4  An anonymous referee points out that Catalan and Occitan also allow bipartite negation, and therefore prefers the hypothesis of a Celtic substratum.While we do not wish to rule this possibility out, it must be noted that in Catalan postverbal pas, like Northern Italian mica, is still a pragmatically constrained optional element, making Catalan (like Northern Italian mica-dialects) a Stage 2 language (cf. Espinal 1993). Occitan does, indeed, have obligatory

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

464   Maj-Britt Mosegaard Hansen & Jacqueline Visconti

There is some evidence that, once French had reached Stage 3, pas, rather than ne, was already felt to be the principal negator, in as much as there are sporadic accounts of ne-deletion among children and lower-class speakers. It is only in the nineteenth century, however, that the language uncontroversially reaches Stage 4 (e.g. Martineau & Mougeon 2003). A number of studies in both real and apparent time have shown that in some varieties of French (see Sankoff & Vincent 1977 for Québecois, Fonseca-Greber 2007 for Swiss French) ne-deletion is now close to categorical in conversational registers, and that it is continually on the increase in France (e.g. Ashby 1981, 2001, Armstrong & Smith 2002, Coveney 1996: ch. 3, Hansen & Malderez 2004, Pohl 1968). One factor that shows a significant correlation with ne-deletion in all these studies is the presence of a clitic subject in the negated clause. Accordingly, several scholars (Harris 1978: 26, Ashby 1981: 681, Armstrong & Smith 2002: 34, Martineau & Mougeon 2003: 140) suggest that the rise in ne-deletion in French from the nineteenth century onwards may have been triggered by increasing fusion of the subject clitics with the finite verb. Martineau & Mougeon (2003: 140ff) in particular adduce a number of syntactic clues suggesting that subject clitics began to move towards affixal status at precisely the time when ne-deletion became noticeable. A potentially confounding factor, however, is the finding, reported in both Ashby (1981: 679) and Coveney (1996: 76f), that preverbal non-subject clitics appear to favor the retention of ne. In addition, Modern French ne is itself a proclitic element, and in principle nothing should prevent sequences of clitics to occur with one and the same verb. Some independent support for the subject-clitic scenario can nevertheless be found in the fact that the Italo-Romance varieties which have progressed beyond Stage 2 are precisely those that are subject to the development of subject clitics. As outlined in Parry (in press), in such dialects there seem to be restrictions on the occurrence of the preverbal negative following the grammaticalization of independent subject pronouns as agreement clitics on the verb. Piedmontese data, for instance, show that if grammaticalized discontinuous structures exist, the preverbal negative is most likely to be dropped when other proclitics occur, thus avoiding the postverbal pas. At the time of writing, however, we have not been able to ascertain how old this phenomenon is, and thus whether it is Occitan which may have influenced French in this respect, or rather the other way around.

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

The evolution of negation in French and Italian    465

more costly operation of substitution (see the Cairese examples in (14)– (16), from Parry 1997): (14) U l’è cin. it it be-PRS-3SG full ‘It is full.’ (15) U n’ è nen cin. it NEG be-PRS-3SG NEG full ‘It is not full.’ (16) U l’ è nen cin. it it be-PRS-3SG NEG full ‘It is not full.’

The ultimate cause of ne-deletion in French is, however, likely to be the fact that, as mentioned above, once bipartite negation becomes established as the canonical form, the function of being the principal semantic exponent of clause negation is gradually transferred from ne to pas, rendering the former redundant. Interestingly, despite the fact that Italian has not reached Stage 3 of Jespersen’s Cycle, the mica-dialects do – albeit to quite a limited extent – allow for non-deletion in substandard (see (17)), and when mica is preposed, also near-standard (see (18)), registers. This fronted, ‘autonomous’ (Cinque 1991 [1976]: 318; see also Visconti 2009: 946f, 2010: 285f) mica deserves further investigation. On the one hand, as stated in grammars of standard Italian, it carries the negative meaning alone (e.g. Serianni 1989: 428).5 On the other hand (as also noted, among others, by Serianni 1989), this use has strong emphatic value. Penello and Pescarini (2008: 51f) suggest that autonomous mica at the beginning of a sentence is different from a negative reinforcer, being rather a ‘probably exclamative’ operator, hosted in the Spec of CP. The clause-initial position, unlike the slot between the finite and the non-finite verb, is indeed a potential focus position.

In this respect it could be viewed as a future stage 6 of the Jespersen scenario (see Table 1, above), following, for example, Ramat (2006a [1994]: 2771; 2006b: 365), who argues, for French based Creole pa and Italian mica, that the possibility of being placed in preverbal position indicates that an NPI has reached full and autonomous negative value. Notice, however, that in medieval French, where they still had NPI uses, pas/mie/point could be placed sentence-initially, but that they gradually lost that ability as they grammaticalized fully as negators (cf. Hansen & Visconti 2009: 146). 5 

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

466   Maj-Britt Mosegaard Hansen & Jacqueline Visconti (17) Ho mica capito. have-PRS-1SG NEG understood-PST.PTCP ‘I haven’t/didn’t [mica] understood/understand.’ (18) Mica (lo) sapevo NEG (it-ACC know-IPFV.PST-1SG che eri sposato. that be-IPFV.PST-2SG married-PST.PTCP ‘I didn’t [mica] know (that) that you were married.’

3.  N-word negation Alongside standard clause negation, both French and Italian have the option of expressing clause negation by means of n-words, that is, expressions which simultaneously display properties of negative quantifiers such as English nobody, nothing, nowhere .  .  . and properties of NPIs such as anybody, anything, anywhere . . . (see Laka Muzarga 1990: 107). With respect to this type of negation in French and Italian, two structures must be considered: (i) those, like (19) and (20), where the n-word occurs postverbally, and (ii) those, like (21) and (22), where an n-word is positioned preverbally. In the former type of construction, French allows deletion of the preverbal marker, whereas Italian does so to a much lesser extent. In the second type of construction, on the other hand, French allows (indeed, normatively demands) insertion of a second preverbal negative marker, whereas Italian does not:6 (19) Je (ne) vois personne. SBJ.CL-1SG (NEG see-PRS-1SG nobody/anybody (20) Non vedo nessuno. NEG see-PRS-1SG nobody ‘I see no-one’/’I don’t see anyone’ (21) Personne (ne) m’ a vu. Nobody (NEG OBJ.CL have-PRS-3SG seen-PST.PTCP Except for some marginal contexts, found in colloquial registers or where a great distance separates the n-word from the verb (see e.g. Manzotti & Rigamonti 1991: 264). When the n-word is topicalized and a prosodic break separates the n-word from the rest of the clause such a construction can result in a double negation (DN) reading (e.g. Corblin & Tovena 2003: 24).

6 

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

The evolution of negation in French and Italian    467 (22) Nessuno (*non) mi ha visto. Nobody (NEG OBJ.CL have-PRS-3SG seen-PST.PTCP ‘No-one has seen/saw me’

Thus, unlike Standard English, French and Italian are both negativeconcord (henceforth, NC) languages, in so far as they allow more than one n-word to occur in a clause without necessarily giving rise to a double (or multiple) negation (henceforth, DN) interpretation. Indeed, in both languages, the default interpretations of (23)–(24) and (26)–(27) will feature only a single instance of negation. Similarly, in clauses such as (19)–(21), above, the combination of an n-word with the standard preverbal negative marker does not result in DN. (23) J(e) (n’) ai rien SBJ.CL-1SG (NEG have-PRS-1SG nothing/anything dit à personne. said-PST.PTCP to nobody/anybody (24) Non ho detto niente NEG have-PRS-1SG said-PST.PTCP nothing/anything a nessuno. to nobody/anybody Default interpretation: ‘I didn’t say anything to anybody’ (25) ?I said nothing to nobody Standard interpretation: ‘I said something to everybody’ (26) Personne (n’) a rien dit. nobody NEG has-PRS-3SG nothing/anything said-PST.PTCP (27) Nessuno ha detto niente. nobody has-PRS-3SG said-PST.PTCP nothing/anything Default interpretation: ‘Nobody said anything’ (28) Nobody said nothing. Standard interpretation: ‘Everybody said something’

However, the extent to which DN is allowed, and, more in general, their NC behavior, varies between the two languages. Starting from a recognition of the most salient differences at the synchronic level, we shall move to a diachronic perspective to propose an explanation. First, the degree to which DN, as opposed to NC, readings are admitted is higher in French than in Italian. Thus, while constructions combin-

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

468   Maj-Britt Mosegaard Hansen & Jacqueline Visconti

ing the standard negator with an n-word can only be interpreted as DN in standard French,7 the NC reading is very strongly preferred in Italian, cf. (29)-(30) vs (31): (29) Je (n)’ ai pas rien dit. SBJ.CL-1SG (NEG have-PRS-1SG NEG nothing said-PST.PTCP (30) Je (n)’ ai pas dit rien. SBJ.CL-1SG (NEG have-PRS-1SG NEG said-PST.PTCP nothing ‘I didn’t say nothing’ = ‘I said something’ (31) Non ho detto niente. NEG have-PRS-1SG said-PST.PTCP anything/nothing ‘I didn’t say anything/I said nothing’

The second difference, which we shall relate to the different origins of French and Italian N-words, concerns the distribution of those n-words in non-negative contexts. Whereas French n-words can be used as NPIs in a number of non-assertive contexts such as interrogatives, conditional antecedents, comparatives etc. (but see further below), non-negative readings of Italian N-words, with the exception of mai, are only clearly acceptable when they occur in postverbal position in yes-no questions, such as (32): (32) C’ è nessuno? here is-PRS-3SG anybody/nobody ‘Is there anybody here?’

A wider range of NPI uses were, however, possible in Old and Medieval Italian, as shown by (33)–(35) (from Parry in press; see also Zanuttini 2010: 576): (33) Doma(n)dà s’ el li vit arma nesuna, Ask-PST-3SG if he-NOM him-DAT see-PST.3SG weapon NPI ‘He asked if he saw any weapon on him at all.’ (14th c. Venetian, Lio Mazor 1312- 14 p.70) (34) Noxe niente a li fantin si li in nadi harm-PRS-3SG NPI to the children if they be-PRS-3PL born-PST.PTCP de no-licito matrimonio? from no-legal marriage? ‘Does it harm children at all, if they are born of an illegal marriage?’ (14th c. Lombard (Mil.), Elucidario, p. 160, 11–12) Québecois is different in this respect: in that dialect, the NC reading would appear to be the preferred one of (29), at least.

7 

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

The evolution of negation in French and Italian    469 (35) Et eo plu tosto lo vorria patere a la mia persone, davante che nullo ambassatore patesse nulla before that NPI ambassador suffer-IPFV.PST.SBJV.3SG NPI vergogna oy offensa inde la corte mia shame or offence in the court my ‘And I would prefer to suffer it personally, before any ambassador suffer any shame or insult in my court.’ (14th c. Neap. Destr. de Troya, p. 130: ll. 4–6)

In French, such uses persist to the present day, albeit to different degrees for different n-words, and on the whole, more commonly in literary than in colloquial registers (although it would be wrong to assume that NPI uses are absent from the latter, cf. Hansen 2011b). Thus, jamais and, to a lesser extent, rien, personne, and aucun, are the most flexible items, the former admitting of positive readings in 11 different non-assertive contexts (see Muller 1991: 265), with plus at the other end of the continuum, positive readings of this adverb being found systematically in only one context (following the preposition/subordinating conjunction sans (que) ‘without’), see (36)–(38), all of which are (very slight adaptations of) twenty-first-century examples found in the Frantext data base.8 (36) Demande- lui si ce train se ask-IMP-2SG DAT.OBJ.CL if this train REFL remettra jamais en marche. put-again-FUT-3SG ever in motion ‘Ask him if this train will ever start moving again.’ (37) Il a refusé de voir personne. SBJ.CL-3SG have-PRS-3SG refused-PST.PCTP to see-INF anybody ‘He refused to see anybody.’ Paradoxically, plus at the same time appears to exclude double negation readings when combined with other n-words, see (i) and (ii). (i) Personne n’ a rien dit. nobody NEG have-3SG anything/nothing said-PST.PTCP ‘Nobody said anything.’ = NC vs ‘Nobody said nothing.’ = DN: ‘Everybody said something.’ (ii) Je ne vois plus personne. SBJ.CL-1SG NEG see-1SG no more/anymore anybody/nobody ‘I don’t see anybody anymore.’ = NC vs *‘I see nobody no more.’ = DN: ‘I see somebody these days (unlike previously).’

8 

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

470   Maj-Britt Mosegaard Hansen & Jacqueline Visconti (38) J’ ai essayé une dernière fois, SBJ.CL-1SG have-PRS-1SG tried-PST.PCTP one last time sans plus croire à ce que je without anymore believe-INF in that which SBJ.CL-1SG faisais. do-IPFV.PST-3SG ‘I tried one last time, without any longer believing in what I was doing.’

In our view, the explanation probably lies in a salient difference between Italian and French n-words, namely that while the former (except for the temporal adverbs mai and più) are etymologically negative, and formally marked as such, the latter – apart from the noun/determiner nul (‘no(ne)’) and the related locative adverbial nulle part (‘nowhere’) – are all etymologically positive. As far as type (i) constructions are concerned, this etymological difference appears to make Italian a negative-concord language from the outset, continuing a pattern also found in Latin, but changing its status from pragmatically marked to unmarked. While Classical Latin grammars prescribe use of one negation only, two negations having affirmative value, as in (39), Molinelli (1988: 40) observes that negative concord was in fact an option not just in colloquial registers, as exemplified in (40), but also as a stylistic device in more elevated registers, to add emphasis, cf. (41): (39) Nemo non videt. (Cicero, Lae 99) no-one NEG sees-PRS-3SG ‘No-one doesn’t see.’ = ‘All see.’ (40) Iura te non nociturum swear-IMP-2SG you-ACC NEG intending-to-harm-FUT.PTCP esse homini nemini be-INF man-DAT no-DAT ‘Swear that you intend to harm no man.’ (Plautus, Miles 1411) (41) Nemini illorum molestus no-one-DAT these-GEN.PL troublesome-NOM nulla in re umquam fuisti no-ABL in thing-ABL ever be-PFV.PST-2SG ‘You never imposed on any of them in any circumstance.’ (Quintus Cicero, Comm. Petit cons. 5, 20)

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

The evolution of negation in French and Italian    471

The construction appears to have been common in Late Latin from the fourth century onwards, where grammars testify to its frequency by criticising it as a ‘mistake’ or ‘grecism’ (e.g. Diomedes, but also Augustinus vs Nonius Marcellus, quoted in Rönsch 1965: 447, who defends the construction as having been used by Varro already).9 As Parry (in press) notes, NC in informal early and late Latin texts of the entire Romance area reflects a tendency to reinforce denial or rejection by negating at both constituent and sentence level, and its continued existence is attested in both early French and early Italian documents, see (42)–(43): (42) Niule cose non la pouret omque pleier no thing NEG her-ACC could-IPFV.PST-3SG ever please-INF ‘Nothing could ever please her.’ (Seq. Ste Eulalie, 880) (43) e’ no soi aizià de doner te nient I-NOM NEG be-PRS-1SG able to give you-OBL nothing aora a questa fiaa. now at this time ‘I cannot give you anything at this moment’. (13th c. Pied.,Sermoni S, 7, 238: 16–17)

Modern Standard Italian, as we have seen above, shows so-called ‘asymmetric’ NC, whereby a postverbal n-word must always occur in the scope of a standard clausal negative or other negative element, but a preverbal one cannot occur with the standard clausal negator (cf. the contrast between (20) and (22), above). In the medieval period, however, many regions seem to have allowed the latter pattern (Molinelli 1988: 55; Zanuttini 2010: 574; Parry, in press), so-called strict NC: (44) che nessum no possa esse prior de la dita that no-one NEG can-PRS-SUBJ-3SG be-INF prior of the said-PST.PTCP Caritay o Confraria no ma[i] de III agni [in]seme. charity or confraternity NEG more than three years [together ‘that no-one may be prior of the said Charity or Confraternity for more than three consecutive years.’ (14th. c. Liguria, Statuti della Compagnia dei Carovana, p. 9, 37–38) 9  Language contact, i.e. influence from Greek, was hypothesized as the source of NC in colloquial Latin as early as 420–30 by Augustinus in his Locutionum in Heptateuchum libri VII. Molinelli (1988: 57ff) argues, however, that its emergence is concomitant with, and more likely to be related to, the expansion of OV word order in the fourth century.

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

472   Maj-Britt Mosegaard Hansen & Jacqueline Visconti Table 2.  The evolution of French n-word negation Stage 1

Je ne dis (rien) ‘I do not say (a thing)’

Stage 2

Je ne dis rien ‘I don’t say anything’ Je (ne) dis rien ‘I don’t say N-word optionally accompanied by anything/I say nothing’ preverbal ne Je dis rien ‘I say nothing’ Negative quantifier)

Stage 3 (Stage 4 [future French?])

A positive NP optionally accompanies preverbal ne to make the scope of the negation explicit ne + Negative Polarity Item

In contrast to Italian, negative concord in Modern French, the case of nul(le part) aside,10 developed gradually out of what was originally a reinforcement of standard clausal negation by positive items with scalar properties.11 This peculiarity, in addition to the development of pas as a postverbal negator (see sect. 1), seems to us to go a long way towards accounting for the differential pace of the grammaticalization of n-word negation in French vs Italian. Thus, we may posit the existence of a ‘quantifier cycle’ that is operative in French alongside the standard Jespersen Cycle, as illustrated in Table  2.12 The table represents a simplified account, however: individual French n-words have substantially different (positive vs negative, nominal vs adverbial) origins, as a result of which their diachronic trajectories are not identical (cf. Hansen 2011b, 2012). Hence, as we have seen above, they exhibit different synchronic properties, in particular the fact that some of them retain positive readings across a much wider range of non-assertive polarity contexts than others. Unlike that of pas, the exact status of French n-words is therefore still unclear, and it is perhaps significant that, in colloquial speech, ne appears to be less frequently deleted in n-word clauses Note, however, that nul(le) part, despite its clearly negative etymology, was also used in “weak” (i.e. non-negative) NPI contexts in medieval French (cf. Ingham 2008). That said, to our knowledge, positive-polarity uses are not attested for nul(le) part. 11  That said, evidence adduced in Hansen (2012) suggests that jamais, etymologically a combination of the adverbs ja (< Lat. IAM ‘(as of) now’) and mais (< Lat. MAGIS ‘more’), was in fact lexicalized as a negative temporal adverb, subsequently developing NPI uses as a result of semantic extension. 12  The evolution of n-word negation in Italian of course bears some resemblance to this cycle. 10 

Brought to you by | The University of Manchester Library Authenticated | [email protected] author's copy Download Date | 11/1/12 11:23 AM

The evolution of negation in French and Italian    473

than in standard negative clauses and that, furthermore, the different quantifiers do not appear to favor ne-deletion to the same extent (Ashby 1981: 678; Coveney 1996: 76; Hansen & Malderez 2004: 23). Nonetheless, it is uncontroversial that the Modern French n-words have been subject to an on-going process of grammaticalization, which is particularly evident in the case of those n-words that are of nominal origin. Thus, for instance, rien (