The Pragmatic Motifs of the Jespersen Cycle: Default, Activation, and the History of Negation in French
The purpose of this article is to delimit the role of pragmatic specialisation in the evolution of negation in French. The change in the marking of sentential negation is believed to proceed in characterised stages that would together constitute the Jespersen cycle. As a marker becomes the default expression of negation, the other markers do not necessarily fade away, and are maintained with specialised roles that include pragmatic functions. One such pragmatic function is that of activation (Dryer 1996), by which a proposition is presented as accessible to the hearer. Activation is shown to motivate the use of preverbal non, which competes with ne for several centuries. The claims that the emergence of postverbal pas in early French and the loss of ne in contemporary spoken French are associated with activation are considered on the basis of novel data. It is concluded that pragmatic functions contribute to language change by providing marked options that may be reanalysed with default status in a grammatical paradigm.
Negation, Jespersen cycle, default, markedness, pragmatic activation, French.
The Pragmatic Motifs of the Jespersen Cycle: Default, Activation, and the History of Negation in French
Since Jespersen (1917),1 the grammar of negation has been assumed to evolve following a cycle of successive stages. The expression of negation goes from a preverbal marker, to the preverbal marker plus a postverbal item, to the postverbal item alone. Well attested across languages, these three2 successive stages3 are generally illustrated by the following constructed realisation of I don't say in Old, Classical and Contemporary French:
(1) Jeo ne dis
I NEG say
(2) Je ne dis pas
I NEG say NEG
(3) Je dis pas
I say NEG
This raises two major questions:
1. Why should stages in the marking of negation emerge and decline?
2. Why should these successive stages form a cycle?
This article primarily concerns the first question. The answer that is generally provided is that the emergence of new markers is motivated by pragmatic functions. The emergence of the stage 2 negative would be motivated by emphasis that would subside with increased frequency of usage (Detges and Waltereit 2002, Eckardt 2006, Hopper and Traugott 2003, Kiparsky and Condoravdi 2006, to cite only the principal recent studies). This is examined in the present article from three perspectives, concerning the status of emphasis and the alternative pragmatic notion of activation (Schwenter 2006), the purported causality of frequency, and the contribution of pragmatics to the relation between marked and default expressions in grammatical change. Issues arise as to the pragmatic notion of emphasis that is often used as a blanket characterisation not clearly related to empirical realisations, and as to whether a better characterisation can be offered by activation. The idea that change in the pragmatic status of a marked expression is brought about by usage frequency variation rather than reanalysis leads to the question of what causes the frequency variation in the first place, and whether this variation is sufficient to lead to the emergence of a new default marker or the decline of an existing marked expression. This brings to the fore the relation between a default expression and its marked alternatives in a grammatical paradigm as a precondition for grammatical change.
The current work has two aims. The first is to adjudicate the thesis that marked expressions of negation are sustained by specialised functions among which pragmatic values are found. The intervention of the pragmatic value of activation is put to the test on the basis of new data. The second one is to document the assumption that changes of stage in the Jespersen cycle are related to changes in the default negative, and that the identity of the default negative defines a stage. The changeable status of items as default or marked in grammatical paradigms is hence a general condition of language change (and plausibly variation). The subtext is that the value of emphasis, the causality of frequency and the view that seeks to marginalise marked options are problematic because they are unclear or unfactual. One thing that this work does not do is to propose an exploration of the critical moments of change, when pragmatic value is gained or lost or when an expression is reanalysed as default or marked, as such a contribution would go well beyond the scope of the current work, which nonetheless offers speculations on the matter in the conclusion. In addition to providing a general contribution to the understanding of variation and change through the relations between default and marked expressions in a grammatical paradigm, my objective is to present an empirically motivated picture of the role of pragmatic activation in the emergence and decline of expressions shaping the Jespersen cycle in French, by showing on the basis of novel data when activation can plausibly be said to intervene and when it cannot. This allows speculations to be made as to the reasons why the successive stages of default negation form a cycle.
These objectives are pursued as follows. The pragmatic value of negatives is examined in the opening section. It considers emphasis, and whether the facts that this label is applied to would be better characterised as activation in the sense of Dryer (1996) and Schwenter (2005, 2006). A pragmatically activated value is one way in which an emerging or declining expression can be marked in its discrete opposition to the default representative of the grammatical paradigm: change in a paradigm is driven by reanalysis rather than gradual increasing or decreasing usage, and reanalysis of a marker as default is what defines a stage in the cycle. The sections that follow consider expressions marked by an activated value on the basis of novel data. It is shown that the decline of preverbal non is associated with a pragmatically activated value in that its distribution is restricted to activated propositions. The claim that activated contexts promote the emergence of the postverbal adverb pas is evaluated by reference to new data closer to everyday Old French usage than literary texts might be. Existing corpora of contemporary conversational French allow for an assessment of whether declining ne is similarly associated with a pragmatic contribution. How these come to form a cycle is speculated upon in the conclusion. The argument is, in summary, that grammatical change should be conceived of as a competition between default and marked forms, and that marked negatives may have an activated value.
2. Changes of grammatical state
The historical grammar of negation has been largely viewed in terms of weakening and strengthening. This perspective has been applied to the well-documented evolution of literary French. Preverbal ne that is the preferred early French sentence negation is formally wearing out: it is a clitic, agglutinates to some vowel-initial clitics and verbs, having no prosodic and syntactic autonomy of its own.4 This state of affairs may threaten the clear perception of negation.5 Clarity as to the polarity of the sentence is brought by the formal intervention of a postverbal minimiser that has an affinity with negative contexts. Minimal measure is expressed in the first instance by items such as pas, point and mie in conjunction with lexically relevant verbs; ne marcher pas would have meant 'not to walk by even so much as a step'. The restriction to negative polarity environments6 may be accompanied by use with an extended set of verbs7 where no reference is made to the meaning of the original noun, as observed by Meillet (1912). From this, the postverbal items develop into negatives8. The embracing negation is initially pragmatically marked. Its contribution would correspond to emphasis whether pas qualifies as a negative polarity item, which has widening and strengthening values (Kadmon and Landman 1993), or a negative proper. Emphasis would be lost with the increased frequency of usage of embracing negation that would in time transform it into the standard negative expression.
Thus, syntagmatic reinforcement would give way to emphatic negation that increased frequency would reduce to a standard negative. This opposition between the emphatic and the neutral negatives proposed by Kiparsky and Condoravdi (2006) can be related to generic communicative requirements of production and perception: a speaker might want to make her position very clear with respect to whether a state of affairs obtains or not, and the hearer might gradually attribute less weight to a frequent structure. This tallies with the dual tension of Larry Horn's neogricean pragmatic theory (Horn 1984 and 1993 among other publications).
This story leaves some questions unanswered, with respect to the nature of the pragmatic notion, to its extension to all elements of the cycle and to the causal role attributed to frequency.
Emphasis is a rather intuitive term. A definition cited by Schwenter (2006: 331) is that proposed by Israel (1998) of the emphatic unilaterally entailing the non-emphatic. Ne marcher pas would thus have entailed ne marcher, but not the other way around. What predictions this makes and how they are to be tested on literary texts of yore is uncertain: it is telling that such uncertainties carry to typological data, and that it cannot be established by van der Auwera's own admission whether many of the data he cites involve emphasis or not. A more testable perspective is proposed by Schwenter through the notion of activation. Schwenter (2005) observes that Brazilian Portuguese postverbal negation não occurs with preverbal não in a discourse-old proposition, which is accessible9 to both speaker and hearer, be it through previous explicit use of the sequence or through inference. Thus, in the following, the use of the postverbal negation with the verb vote alone corresponds to deny voting for a certain candidate, on the strenght of the previous mention of voting for that candidate:
(4) A : O João votou no Lula?
'Did João vote for Lula?'
B1: (Não.) Não votou não.
'(No.) He didn't vote (for him).' (p. 1445, example (17))
Contextual information is indexed because of the activated proposition that is required by the postverbal item. The preverbal negation in the answer to the same question does not suppose an activated proposition and therefore does not index the information from the antecedent sentence; it thus simply denies voting at all:
(5) B2 : (Não.) Ele não votou.10
'(No.) He didn't vote (for anyone).' (p. 1445, example (17))
The contrast indicates that an activated proposition is a necessary condition for the use of the postverbal negation in Brazilian Portuguese. The requirement is shown to characterise postverbal markers pas and mica in Catalan and Italian. An example of this is provided in the following, where it is not disagreement that allows B to use postverbal pas, as B concurs with A and is therefore unlikely to want to emphasise the negative, but the mere fact that the proposition was explicitly used and is therefore accessible to hearer and speaker alike:
(6) A: La Maria ja no vindrà a aquestas hores.
the Maria already not will.come at these hours
'Maria won't be coming at this hour.'
B: Efectivament, la Maria no vindrà pas tan tard.
indeed the Maria not will.come NEG so late
'True, Maria won't come so late.' (Schwenter 2006: 334, example (4))
As the emergence of postverbal markers is a central development of the negative cycle, it may be that it is promoted by activated contexts. That is the hypothesis formulated by Schwenter (2006) and taken up by Mosegaard Hansen and Visconti (2007). The latter seek to establish whether the emergence of a postverbal negative marker in early French and Italian is related to explicit activation. Such cases are attested:
(7) “Se vous me voulés croire, …” Il ne le vorent pas croire, …
“If you’ll believe me, …” They wouldn’t believe him, … (Mosegaard Hansen and Visconti 2007: 10, example (32))
but they are very much the minority case, and most instances involve implicit inferences brought about by lexical relations. The inference of ignorance deriving from not having been told something makes the second proposition in the following early French sequence accessible, which in turns allows the use of postverbal mie.
(8) Ne l'oï dire, ne jo mie nel sai (Roland v. 1386) (Mosegaard Hansen and Visconti: 22, example (101))
I haven't been told, I don't know it [at all]
The notion of activation could well constitute a better characterisation of the contexts promoting the emergence of postverbal items in the negative cycle, if only because it makes testable claims about the distribution of items.
It is not only emerging items that are relevant to the Jespersen cycle, but also declining items. Such declines are illustrated by the preverbal item non in early French and by ne in the contemporary period. Could these declines be correlated to a pragmatic value such as activation? If this were the case, it would bring a reconsideration of the respective roles of pragmatic specialisation and frequency; frequency would not only reduce pragmatic contribution, it may also create it for increasingly lesser used items. It would also force a revision of the claim by Kiparsky and Condoravdi that pragmatic charge can only come from syntagmatic reinforcement, a claim which is already suspect given the fact that they themselves mention that "[y]es and no were originally reserved for emphatic assertion and denial, and supplanted their plain counterpart yea and nay in Middle English" (2006: 3).
Finally, the causal role attributed to frequency changes demands clarification. Increased frequency of usage is argued to turn a pragmatically charged negative into a standard one, and it may be that decreasing frequency turns a standard negator into a specialised one. Two important questions arise as to how both standard and marked negatives can be affected by frequency. First, as it is a gradual concept, if frequency affected the pragmatic import, this would mean that different degrees of emphasis or of activation should be observable; the difficulties in assessing emphasis alone make degrees of it impossible to diagnose; degrees of activation seem in the current state of knowledge to be an enigmatic concept. Secondly, why should frequency suddenly change for pragmatically specialised negators? Those particular values should constitute at any period a small proportion of all negative uses,11 unless there are reasons to propose that speakers of a given period become more emphatic as it were, reasons that we do not have.12 It may however be that specific expressions become more frequent: of the 400 or so minimisers found in early French by Möhren (1980), only pas, point and mie become negatives,13 with pas by far the most common form. Yet, even if pas were the only pragmatically marked form, would its frequency be enough to both diminish its specialised contribution and challenge ne alone as a standard negation? Surely, lack of special pragmatic role and the increased frequency for ne ... pas might well be a consequence rather than a cause of pas becoming the default negative. A rise in frequency cannot in and of itself bring about change of status. As the default negative should be the most frequent expression in the paradigm, accessions to default status should therefore explain rise in frequency. The default expression should provide a neutral pragmatic contribution, as "an obligatory element cannot be emphatic, for to emphasize everything is to emphasize nothing" (Kiparsky and Condoravdi 2006: 5). Obligatory must be taken to mean the default14 expression of sentence negation as opposed to marked items and structures.15
In other words, it is reanalysis as default negation that explains higher frequency, less restricted usage conditions and the absence of specialised pragmatic contribution, rather than the other way around. If this is correct, it means not only that the default structure is pragmatically neutral, but conversely that a specialised pragmatic import correlates to the marked expressions. Such a prediction would apply equally to emerging and declining expressions. The opposition between marked and default expressions would be central to the understanding of the Jespersen cycle and other grammatical changes.
The perspective outlined here not only has the advantage of doing away with degrees of discrete pragmatic categories supposed by a frequency-driven analysis; it also explains a puzzle relating to the competition of forms in language change. Change in grammar does not lead to an immediate change of available forms, which remain in competition. Thus, citing Parry (1997: 244), Floricic notes that "in certain dialects of Val Bormida, the three stages of the Jespersen cycle are synchronically attested, even if a priori the discontinuous negation represents the default case”16 (2005: 169; my translation, PL), as would be the case in Gascon (n. 14, p. 191); van der Auwera (2008) provides further examples of competition between markers and stages in the same language. This can be illustrated by the grammar of French negation and the three essential changes it has witnessed since the first records. The first is the disappearance of non as a preverbal negation where only ne remained. The second is the emergence of postverbal negatives. The last is the marginalisation of ne.17 At least three modes of negating the verb are thus available in Old French, non alone, ne alone18 and ne with a postverbal marker. Modern French retains ne alone, ne with a postverbal marker and the postverbal marker alone. Setting aside the preverbal use of non, this situation could lead to the belief that no change has taken place in the history of French negation. It remains the case ten centuries later that ne alone is still an option for the expression of negation (Larrivée 1995); it can yield focus, license negative polarity items and give way to double negation, as any negative item should (Larrivée 2004: chapter 1). This calls into question the various claims that pas in contemporary French is the only negation, which could just as well be expressed by ne. It is true that this option is specialised in its distribution, both in terms of formal written register and with respect to the averidical contexts in which it occurs (with modals and indefinites, in interrogative and conditional clauses). It has a lower frequency19 than the embracing negation that in the formal register constitutes the immediately available option in all contexts. The contemporary situation is in the end different from early French, the change being that French negation can be expressed not by one form surviving all others, not by the form being the only 'real' negation, but by that form becoming the default option while others are reanalysed as marked. This view is convergent with the discussion by Breitbarth and Haegeman (2008) of why marked preverbal negatives might be maintained:
(9) For a language to maintain such a preverbal marker for an extended period of time, this marker has to acquire further functional specialisation. (Breitbarth and Haegeman 2008: 4)
The proposal is that the changes in the grammar of negation equate to a change in the default marking of the category. The other means of expressing the category must then assume a specialised role. This division of labour between marked and default as a factor informing language change is what is assessed for the three major changes identified in the history of French negation. I more specifically consider when the marked option can be said to be characterised by pragmatic activation, a plausible value for specialised expressions that can be readily tested. This should tell us about the diagnostic of activation on novel data close to the vernacular, the contribution of pragmatics to grammatical paradigms and the central role of the opposition between marked and default expressions in language change. Section 3 looks at the disappearance of non in Old French, section 4 at the emergence of pas in Old French and section 5 at the disappearance of ne in contemporary French.
3. Leaving stage I: Non in early French
The disappearance of preverbal non20 from French negators seems never to have been taken very seriously. Few studies have dealt with it despite the fact that it is the only major innovation in the French grammatical paradigm of negation as opposed to its Romance counterparts. It furthermore contradicts the usual understanding of the Jespersen cycle, which is said to lie in the weakness of ne; it is the strong non that specialises and disappears first, with the last example attested in 170821 (Reid 1939: 305, n. 1). The studies that consider the preverbal marker do so as a paradigmatic comparative to ne in the actual/virtual semantic system of Moignet (1965; see also Guiraud 1964, Martin 1972), and to nient by Taylor (1976). Only Reid (1939)22 offers a characterisation of the distribution of preverbal non modifying a finite verb.23 In direct competition with ne24, the use of preverbal non is characterised as "markedly affective" (p. 306). It is used for "denying the truth of a statement, but also making a negative reply to a question, and refusing to obey a command" (p. 306). It relates mainly to verbs avoir, être, to modals and to faire. The sequence is "always in some degree elliptical, the sentence being meaningless except with reference to what precedes" (p. 306). It is accompanied by "formulas of asseveration such as certes, voir, par foi, par m'ame" (p. 306). These characteristics converge in the following illustration:25
(10) Juré l'avons et fiancié,
Cados, il fix mon oncle, et gié,
Que nos par force te prendrons
Et a mon oncle te rendrons,
Qui moult te het et mout t'a vil.
Por voi te di q'ensi ert il.
'For truth you-ACC say-PS-1PSG that so be-FUT-3PSG it'
– Non ert, se Dix me veit secorre.
'Not be-FUT-3PSG, if God me-DAT wishes help'
(Lancelot, Ille, 530-536; Denoyelle 2007)
"We have sworn and promised,
Cadox, my son's uncle, and I,
That by force we'll take you
And to my uncle give you back
Who very much hates you and holds you very much as vile
In truth I tell you that it will be so
– It will not be so, if God wishes to help me”
The response where non modifies the verb be denies the antecedent assertion it will be so. It is elliptical, the manner complement is missing, and it refers specifically to the antecedent threat. The denial is reinforced by the expression if God wishes to help me.
Such reinforcements are not always present, and the verb involved varies.
(11) Honte i avrai et reproche toz tans
'Shame I have-FUT-1PSG and reproaches all times'
– Non avrez, sire, dist Vivïens li frans. (Aliscans, 204-5)
'Not have-FUT-2PP, Sir, said Viviens the noble-one'
J'aurais pour toujours honte et reproches.
– Non, seigneur, répond le noble Vivien. (Buridant 2000: 704)
"I will forever have shame and reproaches.
– Not so, Sir, answered the noble Vivien."
(12) Dist Chantecler : Renart cousin,
Vos me volez traire a enging?
'You me-ACC want-PR-2PP draw at engine'
– Certes, ce dist Renart, non voil (RenartR, I, 4365-67)
'Certainly, this said Renart, not want-PR-1SG
Chantecler dit: Renart, mon cousin, vous voulez me berner.
– Assurément, dit Renart, ce n'est pas du tout mon intention. (Buridant 2000: 704)
"Chantecler said: Renart, my cousin, you want to fool me?
– It is certainly not at all my intention, said Renart"
(13) Nel ferez/
Par vos li mant qu'il nos soit secoranz
'By you him-DAT ask that he us-DAT be-SUBJ-3PSG rescuing
– Non feré, sire, dist Bertrand li vaillanz.
'Not do-FUT-1SG, Sir, said Bertrand the brave'
Vous lui demanderez de notre part de venir nous secourir.
– Non, seigneur, répliqua Bertrand le valeureux. (Buridant 2000: 704)
"You will ask him to come and rescue us
– I will not do so, Sir, replied Bertrand the brave."
Variation is found also in the speech acts accomplished: apart from the negation of an assertion26 (10), of a question (12), of a request (13), agreement with an antecedent negative assertion, and its extension to other arguments is also attested:
(14) Je vous pri que ne me reffusez pas.
– Non feray-je, m'amie, par ma foy.
(Quinze Joye de mariage, v; Reid 1939: 309)
"I beg you that you do not refuse me.
– I will not, my dear, by my faith."
(15) Le musnier fut content, et jamais plus n'en parla; non fist le seigneur que je sache (Cent Nouvelles Nouvelles, iii; Reid 1939: 308)
"The miller was satisfied, and never spoke anymore of it; the lord didn't either as far as I know"
What does not vary, however, is the dependence on an antecedent proposition that is overtly accessible. In other words, explicitly activated propositions license the use of preverbal non, that is not itself activated but used in a proposition that is.27 The dependence on an antecedent proposition is shown by the information inherited from it by the elliptical sequence in which non is used.28 That sequence can thus confirm the antecedent proposition, and not just contradict it: noted in the case of the Catalan example in (6), this is observed in the two attestations (14) and (15) above, and casts further doubt that emphasis could be involved as emphatic negation seems to be associated with disagreement. Contextual dependency applies to the two illustrations set aside in the Anglo-Norman dictionary. The first is a straightforward denial of the explicitly activated proposition you know what he said:
(16) Respondi Hieu: 'Bien le conoissiés et ce k'il a parlé savés'. Et il disent: 'Non savom' (Liv Reis 191; also Reid 1939: 307)
"Hieu responded: 'You knew him well and you know what he said.' And they said: 'We do not."
The second does not seem to contain explicit activation, although the relationship between the evoked threat and the request for help qualifies as a case of inferred activation, as a help request can be expected in the context of danger:
(17) sont en poynt d'estre anientisez si noun soit par le graciouse aide [...] de vostre [...] seignurie (TextLett and Pet 75.46)
'are on point of be-INF annihilated if not be-SUBJ-3PSG by the gracious help of your Lordship'
"[They] are on the verge of being annihilated but for the gracious help of your Lordship"
It may be that this exemplifies a different structure such as the focusing one (mentioned in note 23), or possibly a case of implicit activation, the only one attested in the consulted resources.
The use of preverbal non thus relies on activated propositions, but not all activated propositions force the use of non, where ne can be used, especially from the 16th century (Reid 1939: 310-311):
(18) Li lous crie: Tu me menaces!
'it-ACC him-DAT shout-PR-3PSG: You-NOM me-ACC threatens
– Ne fes, sire, salvez vos graces.
'Not do-PR-1SG, Sir, save-IMP-2PP your graces'
(Ysopet de Lyon, ed. Foerster, 84)
"He shouts to him: you're threatening me!
– I do not, Sir, save your graces"
This illustrates the point made by Breitbarth and Haegeman (2008) that special pragmatic contributions do not have to be signalled categorically. Optionality of pragmatic markers makes it possible for changes in default and marked status of existing expressions that are considerably simpler than wholesale semiological modifications. Several expressions can contrive to indicate a pragmatic value, and non can be accompanied by pas or nient as early as in the 12th century (Reid 1939: 309):
(19) Ce sont bien les vostres, dit-il.
'It be-PR-3PP indeed the yours, say-PstHist-3SG-he
– Les nostres! Non sont pas. (Des Perriers, Récréations, iii)
'The ours! Not be-PR-3PP not'
"These are indeed yours, said he
– Ours! Not so!
This of course supposes that pas would be marked for activation. Whether the markedness of emerging postverbal marker pas does indeed have a pragmatic import is considered in the next section.