The distribution of non, nenny and non fait in Pre-Classical and Classical French

Abstract

The aim of this study is to track the evolution in the use of the markers nenny, non + verb (non fait ‘no, it doesn’t’) and non in its absolute use between the middle of the 15^th century and the end of the 18^th. In Middle French, non already covers all the uses of the old markers nenny and non fait, but it remains in the minority. In Pre-Classical French (1550–1650), the frequency of nenny and non fait decreases considerably and, in Classical French (1650–1789), they become archaic. In the mid-17^th century, non definitively assumes the functions of the medieval markers, which disappeared. The analysis of the temporal distribution of these markers helps to date the transition from ancient to modern uses. Several studies of phonetic, morphological and syntactic phenomena have also aimed to date the turning point between the medieval and the “classical” language, which occurs during the so-called “pre-classical” period. This research also seeks to contribute to the debate on the position of the boundary between Pre-Classical and Classical French on the basis of pragmatic criteria. The results support placing this boundary within the decade 1620–1630, as other studies did for morphosyntactic phenomena.

Résumé

L’objectif de cette étude est de tracer l’évolution de l’emploi des marqueurs nenny, non + verbe (non fait) et non en emploi absolu entre le milieu du XV^e siècle et la fin du XVIII^e. En moyen français, non recouvre déjà tous les usages des marqueurs anciens nenny et non fait, mais il reste minoritaire. En français préclassique (1550–1650), la fréquence de nenny et de non fait diminue considérablement et, en français classique (1650–1789), ils deviennent archaïques. Au milieu du XVII^e siècle, non assume définitivement les fonctions des marqueurs médiévaux, qui disparaissent. L’analyse de la distribution temporelle de ces marqueurs permet de dater la transition entre les usages anciens et les modernes. Nombre d’études portant sur des phénomènes phonétiques, morphologiques et syntaxiques ont cherché aussi à dater le tournant entre la langue médiévale et la langue « classique », qui a lieu pendant la période dite « pré-classique ». Cette recherche veut contribuer également à la réflexion sur la position de la frontière entre le français pré-classique et classique par des critères pragmatiques. Les résultats permettent de situer cette frontière plutôt dans la décennie 1620–1630, comme d’autres recherches l’ont fait des pour des phénomènes morphosyntaxiques.

1. INTRODUCTION

In this article we analyse the evolution of the distribution of the response particles nenny and non, and the disagreement responsive structure non followed by a finite verb – non feray (‘I won’t do this’), non est (‘it’s not’), non a (‘he hasn’t’) –, in drama texts from the mid-15^th to the late 17^th century. The mid-17^th century is considered a turning point in the history of French, as it corresponds to the culmination and the beginning of several morphosyntactic changes. For linguists such as Combettes (2003) and Combettes and Marchello-Nizia (2008), this turn describes the transition from Pre-Classical to Classical French. More recently, new studies (Ayres-Bennett and Caron, 2016; Amatuzzi et al., 2020) have examined the dating of this delimitation, proposing new dates and new perspectives on the periodization of French. An additional aim is to contribute to the discussion regarding the boundary between Pre-Classical and Classical French by analysing pragmatic data – the use of the (dis)agreement markers nenny, non and non + V –, instead of morphosyntactic data, which is more usual in this type of study.

The question of periodization underlies any research in diachrony. In a more or less conventional way, linguists describe a specific phenomenon placing it within predefined time boundaries. ¹ There is a broad consensus regarding the division of the history of French into five stages: Old French (842–1330), Middle French (1330–1550), Pre-Classical French (1550–1650), Classical French (1650–1789) and Contemporary French (19^th and 20^th centuries). ² However, as Combettes and Marchello-Nizia (2008: 355) point out, there is nothing immutable about periodization. Indeed, advances in the description of the functioning of the language throughout history make it possible to shift boundaries, establish new ones or remove them altogether.

The segmentation criteria of French are usually based on phonological, morphological, syntactic or lexical phenomena. To the best of our knowledge, there is a regrettable lack of works on periodization based on the study of pragmatic markers ³ (Gómez-Jordana, 2013). In this way, we will evaluate the accuracy of the commonly accepted final boundary, namely 1650, from a pragmatic and conversational perspective but also reflect on the relevance of establishing a “Pre-Classical” period in the history of French, through the study of the evolution of the minimal answers non and nenny ⁴ and the next to minimal answer (Hakulinen, 2001) non + V (non feray, non est, non avra…) from 1450 until the end of the 17^th century.

This research is based on a corpus of 865 tokens from Frantext. The qualitative and quantitative analysis conducted will allow us to identify 1. the discourse contexts in which the (dis)agreement markers non, non + V and nenny are used in Middle French, 2. the evolution in the use of each marker, and 3. the dates on which the uses disappear or increase. The results of this study will be compared with those of Ayres-Bennett and Caron (2016) and Amatuzzi et al. (2020), who identify a break around 1620–1630, rather than 1650. These authors invite further research on other linguistic phenomena in order to confirm their results. This research follows on from theirs considering the role of the three negative markers.

In Old and Middle French, nennil/nenny makes the pair with oïl/ouy, ⁵ and non + V is one of the members of the structure si/non + V (+ pronominal subject) (si suis, non fait, si arai ge, non fera il… ‘I am’, ‘it doesn’t’, ‘I’ll have’, ‘he won’t do it’). In general terms, the difference in the use of the (dis)agreement markers oïl/nennil and si/non + V (+ pronominal subject) is based on the type of utterance they respond to: the pair oïl/nennil is used to answer “real” questions, ⁶ i.e. questions that address a request for information where the speaker who asks it has no prior idea about the answer (or presents it as such). In (1), nennil rejects or disagrees with the proposition of the question sont tout li enfant Rose mort et a fin venu? (‘did all the children of Rose die?’):

The pair si/non + V (+ pronominal subject) is used mainly to disagree with injunctions and assertions, but also with orientation questions, i.e. requests for confirmation of a point of view asserted more or less explicitly by the speaker. Negative interrogatives, for example, are a type of orientation question: the speaker is already directing his question towards a positive or a negative answer (Heritage, 2002). In (2), non ferons disagrees with a request, Yssiez, je vous en prie (‘I beg you to get out of here’), the most frequent type of sequence in Middle French:

The detail of these and other uses of nenny and non + V will be explained in Section 3. The complementary use of the two pairs is well established until the end of Middle French, when non in its absolute use – which already exists in Medieval French but is in the minority – becomes much more frequent leading to the decline of nenny and non + V.

This article is structured as follows: Section 2 reflects on how to approach the periodization of Pre-Classical French, which will help us to define some hypotheses. Then, Section 3 describes the uses of nenny, non + V and bare non between 1450 and 1700. Section 4 studies the distribution of these markers during this period on the basis of a corpus of different texts in different genres for Middle French, as well as in purely dramatic texts from 1550 onwards. Finally, Section 5 summarizes the results of the analysis and presents the general conclusions.

2. HOMOGENEITY AND DELIMITATION: THE QUESTION OF PRE-CLASSICAL FRENCH

Periodization roughly consists in establishing boundaries within an object whose evolution is continuous, and as such is problematic: “délimiter des périodes conduit inévitablement à segmenter une réalité caractérisée en fait par la continuité” (‘delimiting periods inevitably leads to segmenting a reality characterised in fact by continuity’) (Combettes and Marchello-Nizia, 2010: 133). In this incessant evolution, Combettes and Marchello-Nizia (2010) distinguish stability periods and instability phases. ⁸ In both cases, changes occur, but in the stability phases there is an impression of equilibrium, and in the instability phases, an impression of a boiling system. ⁹ Combettes and Marchello-Nizia (2010) also explain that Middle French and Pre-Classical French are unstable periods, which make the transition between more homogeneous or stable language states (Old French and Classical French, in this case).

The periodization of Pre-Classical French is relatively recent compared to the other more or less accepted periods (Medieval French, Classical French and Modern French). Before the isolation of the decades 1550–1650, grammars that described this period in the history of French tended to homogenize the “language of the 17^th century” (Gougenheim, 1974) and the “language of the Renaissance” for the earlier decades (Huchon, 1998). The historical division of this period was based, at best, on historical facts related to the French language and the construction of a norm (the publication of the first grammars from 1520 onwards, the creation of the Académie Française in 1635, etc.), or simply on a temporal delimitation by centuries, which was much more practical.

In the 1990s, broader studies on the history of French (Picoche and Marchello-Nizia, 1991; Marchello-Nizia, 1999) and more specific ones on the French language of the 16^th and 17^th centuries (Buridant, 1997) attempted to characterize the language of this period. These studies already noted – or at least perceived – that the French language of the early 16^th century did not function like the language of the late 16^th and 17^th centuries. There was no linguistic homogeneity within these two centuries. It was a “hybrid” period where old and new usages co-existed. Combettes’s (2003) publication, based on an analysis of syntactic phenomena, implied the introduction of a “pre-classical” stage in the history of French between 1500 and 1650 (or 1660). A few years later, Combettes and Marchello-Nizia (2010) postponed the initial boundary to 1550, on the basis of new linguistic descriptions.

From a linguistic point of view, the pre-classical stage is characterized, among many other features, ¹⁰ by the regularization of the demonstrative determiners (base cet-) and the paradigms of demonstrative pronouns (base cil- and cel-), which alternated in Medieval French. The specialization of the pronouns occurs around 1550, while that of the determiners occurs around 1650, the initial and final dates for Pre-Classical French proposed by Combettes and Marchello-Nizia (2010). During this changing period between the 15^th and 17^th centuries, the sentence order SVO was imposed, which had been limited to VO since the 13^th century. ¹¹ As regards the lexicon, certain alternations of the old language were reduced or fixed, as the adverbs très/beaucoup, whose modern use was established at the expense of moult, or avant/ains, whose old term ains disappeared at the end of the 16^th century.

For authors such as Caron (2002) or Ayres-Bennett and Caron (2016: 344) the notion of “instability period” is not compatible with the very notion of “period”, since the latter presupposes a homogeneity absent in the notion of “instability”. One way of getting around the problem of segmenting a moving object is to use the notion of “chronolect”, analogous to that of “dialect” from a diachronic perspective. Thus, relatively brief moments (limited to about thirty years or one generation) could be distinguished where very rapid changes occur which give rise to a new widespread use in a linguistic community. Contrary to what we might think, changes are not necessarily gradual. There is not always a balance between the old usages and the new ones. Sometimes the change is “dramatic”: after a long decline of the old variant, the new variant clearly takes over (Caron, 2002). However, Traugott (2022: 36–37) distinguishes innovations and changes: innovations are new individual uses of the language that may diffuse (or not) into one or more linguistic communities, giving rise to a change. For her, morphosyntactic change is gradual in the sense that the old and the new uses may co-exist for a long time. This rejects the “catastrophic” hypothesis of the language change. In the lexical domain, the innovations are instantaneous, but their conventionalization process is gradual as well. As Traugott (2022) explains, linguistic change is a matter of language use, so any description of the evolution of language should be based in the analysis of the use of individual speakers in order to determine whether or not a simple innovation spreads leading to an effective change in the language. Thus, a comparison of the frequency of different structures’ use (with their different values) provides evidence on the pace of the change process.

Coming back to the Pre-Classical French’s periodization, Ayres-Bennett and Caron (2016) and Amatuzzi et al. (2020) discuss the accuracy of the boundaries proposed by Combettes and Marchello-Nizia (1550–1650), and even the relevance of defining this period characterized by heterogeneity and constant asynchronous changes. It is obvious that the dates used for any periodization are somewhat arbitrary, on the one hand, because changes do not occur overnight, and on the other hand, because the beginning and the end of each evolutionary process do not coincide with those of another evolutionary process. Moreover, the material linguistic traces that attest to the evolution of the language in written texts – mostly of a literary nature – can not be synchronized with real spoken uses. In other words, the literary written discourse is likely to be more conservative than the spontaneous spoken language and innovations will thus be attested later. ¹² Research into texts belonging to discourse traditions ¹³ that are not written and literary can provide new insights into the division of language history (Kabatek, 2012: 40). That is what Amatuzzi et al. (2020) have done by looking at sermons and correspondence from the 16^th and 17^th centuries. ¹⁴

For Caron (2002), the periodization of Pre-Classical French remains inappropriate or somewhat uncertain. In the continuous evolution of French, Caron observes a break in the 1620s, and Ayres-Bennett and Caron (2016) and Amatuzzi et al. (2020) confirm a very abrupt transition phase around 1630. Among the morphosyntactic innovations described in these two studies are the reduction of allomorphic bases of verbal paradigms (véquit/vécut), the specialization as pronouns or determiners of ambivalent morphemes (chacun/chacune), the rise of clitic pronouns between the verb and the infinitive (je le veux faire/je veux le faire), or the classification either as adverbs or prepositions of certain terms (dedans, dessous, dessus, dehors). All these morphosyntactic evolutionary processes go through a turning point in the 1620s, either because the progressive changes accelerate by selecting the modern usage, or because the old usages start to fall into decline.

Our study on the evolution of the use of the (dis)agreement markers non, nenny and non + estre/avoir/faire does not seek to describe a more relevant change than the morphosyntactic ones mentioned above. Aware that the evolution of language is not continuous at any level (morphological, syntactic, lexical, semantic, pragmatic), our aim is to provide arguments based on the change of a pragmatic feature – using a specific corpus – in order to support Combettes and Marchello-Nizia’s (2010) periodization or Ayres-Bennett and Caron’s (2016) one.

The following section presents the evolution of the use of the Medieval French markers nenny and non + estre/avoir/faire, which disappear in the course of the 17^th century. ¹⁵ We will identify the moment when non in its absolute use assumes the role of the former markers. The aim is to determine whether the years 1620–1630 also represent a turning point at a pragmatic level – at least as far as the expression of disagreement is concerned.

3. DISTRIBUTION OF NENNY, NON + V AND NON (1450-1700)

3.1 Presentation of the data used

As we have already seen, in Ayres-Bennett and Caron’s (2016) and Amatuzzi et al.’s (2020) studies the selection of the data is of great importance from a methodological point of view. The authors analyse texts that would reflect a usage as close as possible to speech, not as a phonic realization, but rather in the conceptual perspective ¹⁶ of Koch and Oesterreicher (2001). Thus, these studies concern sermons and correspondence. In this research we focus exclusively on the dramatic texts ¹⁷ of Frantext dated between 1550 and 1700 in order to analyse the evolution in the use of non, nenny and non + V. Although the authors of classical plays had strict rules of composition (rules of decorum, unity of time, etc.), the language had to be accessible to a contemporary audience: ¹⁸ “[Play-texts] have enough points of similarity to suggest to the audience that the use of knowledge of conversation in their interpretative procedures is appropriate (…). And playwrights exploit this knowledge” (Culpeper and Kytö, 2010: 211). Our corpus from 1550 to 1700 includes 34 texts classified as tragedies, 56 as comedies and 24 as tragicomedies – some of these latter two are humanist or pastoral. ¹⁹

Data in Table 1 do not show a correlation between the drama sub-genre and the use of the old markers (nenny and non + V) or the modern ones (non and its reduplicated version non non). As the three major drama sub-genres (tragedy, comedy and tragicomedy) are present in our corpus – comedy being the most represented one (46%) –, we consider that the representation of the oral language is sufficiently close to the spontaneous usages of the time, at least as regards the refuting markers of this study.

Table 1. Textual sub-genres of the dramatic texts (1550–1700)

In order to analyse the relevance of isolating the “Pre-Classical” period in the history of the French language, data from Frantext of the period 1330–1549 (Middle French) has also been added. Thus, the evolution of the use of nenny, non + estre/avoir/faire and non will be described from the late medieval period. Concerning the second boundary of Pre-Classical French, the extension of the analysis until 1700 will enable us to confirm the trends perceived between 1550 and 1650 for Classical French. The final aim is to determine whether the break described for morphosyntactic structures between 1620 and 1630 also affects the use of disagreement markers nenny, non + estre/avoir/faire and bare non.

Table 2 shows the dimensions of our corpus (number of texts, authors and words), divided in time slots, that are not homogeneous, but neither is the number of texts for each segment nor the number of authors or tokens. Since our aim is to define the transition of Pre-Classical French between 1620 and 1650, wider breaks have been made in Middle French (between 20 and 50 years) and Classical French (between 15 and 25 years). For Pre-Classical French (1550–1659), the time cuts have been set at periods of 10 years, except from 1550 to 1599. The slot 1610–1619 does not count any token or text. ²⁰

Table 2. Dimensions of the corpus analysed

Data from the period 1330–1549 (Middle French) concern all types of text in Frantext and not only the dramatic ones, as for the later states of language (Pre-Classical and Classical French). In previous studies (Saiz-Sánchez, 2016a; 2020a) we found that the pragmatic functioning of nenny and non + estre/avoir/faire is stable in Old and Middle French: they appear in the same contexts in all the textual genres studied (verse and prose novels, chansons de geste, plays, criminal registers, etc.). The only particularity related to the textual genre is the distribution of the structure si/non + V (+ pronominal subject), which mostly expresses disagreement with assertions and orientation questions, but in prose novels of the 13^th century it is much more used to agree with injunctions (Saiz-Sánchez, 2020a: 277-sq). Therefore, considering all textual genres in Medieval French enables us to increase our corpus and to offer a more accurate description of the initial values of our markers. ²¹

This article will focus on the evolution in Pre-Classical French of nenny and non + V, the Old and Middle French disagreement markers. These two markers were the negative terms of the double marker system oïl/nennil (‘yes/no’) and si/non + verb (+ pronominal subject) (si suis, non fait, si aurai ge, non fera il… ‘I am’, ‘it doesn’t’, ‘I will have’, ‘he won’t do it’). All allomorphic variants of nenny (nennil, nennil, nanil, etc.) were taken in account. ²² Concerning the markers of the type non + V, all tokens of non followed by the finite verbs estre, avoir or faire were analysed. ²³

As we shall see later, the drop of use of non + V begins around 1450 and it is non in absolute use that takes over in the 17^th century. However, bare non is already attested as a disagreement marker in the medieval language (Hansen, 2020). The frequency of non as a negation adverb is extremely high because of the other contexts where it may appear to negate the following proposition – or particle – instead of referring to the preceding discourse (i.e. Non seulement il étudie, mais il travaille aussi ‘Not only does he study, but he also works’). To avoid the manual sorting of an overwhelming number of tokens of non, we searched in Frantext for punctuation marks followed by between zero and four tokens and then by non.

The results of this research are shown in Table 3, in which the three last lines correspond to the detail of the non + V structure with the verbs faire, estre and faire.

Table 3. Number of tokens of nenny, non + V and non in Middle, Pre-Classical and Classical French

The general trend is evident: the disagreement markers nenny and non + V of Medieval French decline in Pre-Classical French in favour of non, as shown by Pohl (1976: 206) and more recently by Saiz-Sánchez (2016a; 2020a) and Hansen (2020). A more detailed analysis will be conducted below in Section 3.3 with narrower periods for the purpose of determining the change of pace as well as any eventual turning points.

3.2 Uses of nenny, non + V and non (1450–1700)

This section summarises the pragmatic functioning of nenny, non + estre/avoir/faire and bare non in Middle French (Saiz-Sánchez, 2016a; 2020a; Hansen, 2020). As explained before, only the markers that fit in a dialogue or a dialogical monologue were included in the corpus. ²⁴ In this period, the expressions non + estre/avoir/faire (+ pronominal subject) are preferred to express disagreement. Agreement is generally marked by using oïl (‘yes’), volentiers (‘of course’) or any other confirmatory expression as je le ferai (‘I will do it’), as shown in Saiz-Sánchez (2016b). This description will highlight the innovations in the usage of nenny, non + V and non in Pre-Classical French, from 1550 onwards.

3.2.1 Functions of nennil

In Old and Middle French, the particle nennil/nenny is paired with oïl/ouy and is mostly a negative response to a neutral question that performs a request for information. In (3), the two nenny answer questions (Esperez vous envers moy de mesprendre? and Me voulez vous jouer de passe passe?) that the addressee interprets as non-oriented, i.e. as information requests:

In (4), the interrogative Je ne say si me mentira/De sa promesse (‘I don’t know if his promise is fake’) may be interpreted as an indirect question introduced by the verb savoir in a negative form, je ne say si (‘I don’t know if’):

The clause Je ne say si me mentira/De sa promesse may be interpreted also as an assertion of the speaker’s doubt. Indeed, nenny can also respond to an assertion. In example in (5) nenny follows je te retiens mon gouverneur. The second speaker refutes the assertion and corrects it:

In early Middle French, nennil may also respond to an injunction, but only if it has a negative polarity (at least in this data). In these infrequent cases, nennil is an agreement marker, not a disagreement one. In (6), nenil is used to agree with ne me lais point mener (‘do not let me be taken away’). In (7), nennil responds to gardez que vous ne laissiez point ceste sente (‘do not leave this path’). Although the main clause has positive polarity, at a pragmatical level, nennil agrees with the negative orientation of the utterance:

In the 16^th century, nenny can also mark disagreement with a positive polarity injunction, an innovation compared to Old French. In (8), nenny responds to the order donnez moy ving quatre deniers et allez (‘give me twenty deniers and leave’):

In Medieval French, the (dis)agreement markers rarely appear in isolation. Normally, they are accompanied by a term of address or by an adverb that modalises the assertion such as certes, bien, voir or voirement (Denoyelle, 2007: 3–8). Frantext shows 19 examples between 1357 and 1597 of the combination nenny(,) non, where non reinforces nenny. Non followed by nenny is not attested. The combination nenny(,) non always responds to assertions that are refuted or to “real” questions:

In (9), the speakers are not quite sure where they are. The second asks the first suggesting various places, including a pewter shop. The first speaker responds to the request for information with nenny, non. In (10), nanil, non responds to the assertive utterance Se vous avez de li mestier, querre l’iray (‘If you need her, I’ll ask her to come’) that accomplishes an offer. This offer implies a question, “do you want me to ask her to come?”, which is a request of information as well. In this view, the negative answer nanil is absolutely natural in Medieval French. In all the occurrences, non is optional and reinforces the negative particle nenny.

The cumulation of nenny and non is quite possible in Middle French. In contrast, it is not possible to find the combination of non + V and nenny, since their functions are complementary from a pragmatic point of view: nenny provides new information to the addressee (as it answers a “real” question), whereas non + estre/avoir/faire disagrees with the pre-existing opinion of the first speaker concerning the question he formulates (which is presented as an orientation question) or responds to an injunction.

3.2.2 Functions of non + V

In Medieval French, the refutative expressions non + estre/avoir/faire (non fist, non suis, non a… ‘he didn’t’, ‘I’m not’, ‘he hasn’t’) ²⁵ are part of the system of markers si/non + V (+ pronominal subject), which in Medieval French mark agreement or disagreement, depending on whether the pronominal subject is expressed or not, respectively. The markers studied here only express negative disagreement as the pronominal subject is always omitted. In (11), non fait il agrees with the negative injunction ne vous desplaise; the pronominal subject is not omitted:

The verbs estre, avoir and faire function as pro-forms that replace a finite verb from the previous discourse. In Old French, there was a wider variety of verbs for these expressions. In addition to faire, the auxiliaries estre and avoir, and some modal verbs such as pooir (‘can’), would also enter these constructions. In Pre-Classical French we rarely find verbs other than faire. The positive expressions si + V and the negative ones non + V express disagreement and have divergent evolutions. Si + faire is used up until the Classical French period, frozen into the form si fait, and later reduced to the current contradiction marker si. On the other hand, as we shall see here, non + V dies out towards the beginning of the 17^th century in favour of non in absolute use.

The disagreement markers non + estre/avoir/faire are used to refute positive assertions (12, vous vous truffez de moy), argumentative questions oriented towards a positive response (13, Ne sont mie?), positive injunctions (14, Venés vous en avesque moy) or assertions that accomplish an injunctive act (15, j’en aurés XXV, i.e. give me 25 candles):

In Middle French, the verb can be elided when introduced by a reporting verb + que structure. A reporting verb + que si expression is also often uttered by the other speaker, mostly in the context of a dispute in literary or drama texts. ²⁶ The drop of the verb reflects the process of disintegration of the medieval system si/non + estre/avoir/faire (+ pronominal subject): ²⁷

Example (16) is a case of elision of the verb (faire). The presence of the equivalent positive expression je gage que si just after je gage que non, without the verb faire either, is a sign of the bare use of non when introduced by a reporting verb. On the other hand, in (17), non answers a question that requests an information (do you know Ulespiegle?), a type of utterance that cannot be followed by non + V. The absence of a refutative expression such as reporting verb + que si in the context is normal as it does not correspond to a non + V structure where the verb is elided. Already in Old and Middle French, non could appear in absolute use (Hansen, 2020: 323). In our literary and drama corpus, non is less frequent (19 tokens) than nenny (462 tokens) and non + V (189 tokens). Other text genres document a higher frequency of non in its absolute use, i.e. the criminal registers (see footnote 26).

3.2.3 Functions of non in absolute use

The bare particle non is excluded from the double marker system of Old French ²⁸ formed by oïl/nennil and si/non + V (+ pronominal subject). In Middle French, non in absolute use marks agreement with a negative utterance or disagreement with a positive one. The limited number of tokens of non in the data (19 tokens) for the period 1330–1549 – excluding the reporting verb + que non structure – makes it difficult to describe a general trend. In this period, bare non is mostly introduced in an indirect speech: Frantext shows 140 occurrences of dire/respondre que non (‘to say/to answer no’) and only 5 occurrences of dire/respondre que nennil. As said in 3.3.2, reported speech seems to be the context where bare non begins its expansion.

Outside indirect speech, non appears in dialogues (63%) or in dialogical monologues (37%). In 32% of the cases, non confirms the negative orientation of a question:

In (18), the questions are presented as confirmation requests of the negative assertions “you are not wiser than Plato” and “you are not more constant than Socrates or Cato”. In the answer – uttered by the same speaker – non agrees with these inferred negative assertions, and certes reinforces the assertive value of non. Nennil confirms afterwards the answer. In two cases (10%), non follows the question tag non? that requests also the confirmation of a previous negative assertion (‘I won’t bring him back and neither will you’):

In two other occurrences, the previous question concerns the locutionary act itself, i.e. the accuracy of the chosen term in the utterance. In (20), non returns to the term autrement and confirms the negative assertion Pas ne vauroye par souhait/Que il fust de my autrement (‘I would not want it to be otherwise for me’):

In example (21), non concerns the accuracy of the verb porterons (‘we will take’), to which the speaker opposes the verb trainnerons (‘we will drag’).

Non may also refute a previous positive assertion (21%). In (22), non pas disagrees with je le ferés venir tout incontinant qu’il serait jour et l’amenerés ceens:

In Middle French, the data also show two examples of non as an acceptance marker of a negative injunction, as in (23) ne vous desplaise, or of a negative assertion (Il ne fault point que l’en la chace), which performs an indirect speech act, as in (24):

The expression non pas in (23) agrees with the interdiction ne vous desplaise (‘if this does not displease you’), which is made explicit later by mais je suis bien aise (‘I am pleased’). In (24), the negative assertion il ne faut pas que l’en la chace (‘we must not chase her away”) is also interpreted as a negative injunction to the addressee of the type “do not let her go”, with which non voir agrees. As in 90% of the attestations in this period, non is accompanied by a modal adverb, in (23) pas and in (24) voir. Non is combined with other modal assertion terms such as certes (32%) or dea/dya (10%), ²⁹ or with the negative marker pas (32%).

Between 1330 and 1599, non marks agreement and disagreement, but its use until 1626 is minor (19 occ. outside indirect speech) compared to non + V (202 occ.) or to nenny (479 occ.). From that date onwards, its use increases at the expense of the old markers, which had been in gradual decline. From the second quarter of the 17^th century ³⁰ onwards, dramatic texts show new contexts where non can appear in absolute use, often accompanied by an interjection (25), a term of address (26), or a phrase that explicates its refutative value (27) – in a dialogic monologue:

In (25) and (26), non disagrees with negative interrogations: ne te mocques-tu point? (‘aren’t you making fun of me ?’) and n’as-tu point ici vu deux cavaliers aux coups? (‘haven’t you seen here two knights fighting?’). Both interrogations are orientated towards a positive response, which would have been refuted with non + V in Medieval French (non ai and non est, respectively). In (27), the first two non answer requests of information, suis-je vive/morte? (‘am I alive/dead ?’), which in Medieval French would have been answered by nennil. The third non follows to the interrogative seroit-ce point peut-estre ceste seconde vie dont parlent nos Druydes? (‘might this be the second life our druids talk about?’), which may be interpreted as an orientation question. Therefore, in Medieval French, the answer would have been non est.

The uses of nenny and non + V remain the same until they both disappear. ³¹ During this period of decline, non + V undergoes innovations that detach it from the Old French system si/non + V (+ pronominal subject). The following section looks at the evolution of the distribution of the three negative markers presented. The distribution of their occurrences will allow us to describe how old usages disappear and how modern ones are established.

4. DISTRIBUTION AND EVOLUTION OF NENNY, NON + V AND NON

In this section, the qualitative study that precedes will be complemented by a quantitative analysis. Table 4 reflects the number of tokens in Frantext of the different types of markers presented above. We have added 290 occurrences of the reduplicated marker non non which frequency raises from 1556 onwards. ³²

Table 4. Distribution of non + V, nenny, non, and non non between 1330 and 1700

The disappearance of the old markers nenny and non + V occurs between the end of the 16^th century and the very beginning of the 17^th century. In the period 1610–1619, Frantext shows no occurrences of our markers. After 1610, only two occurrences of non + V were found (in 1633 and 1669). As far as nenny is concerned, data show 14 occurrences between 1550 and 1579 and only three occurrences between 1580 and 1585, then none until 1630. Thereafter, between 1630 and 1700, we found only eight nenny in the 174 plays of Frantext, which is an almost negligible number. Moreover, in these occurrences, the use of nenny is deliberately archaic or shows a lower register. ³³ The absolute use of non – not introduced by a reporting verb + que – is infrequent until 1620-1629, when there is a notable peak in frequency (see Figure 1). Between 1330 and 1599 we find only 19 occurrences of non in 373 texts, then, between 1600 and 1625, no occurrences are found. Afterwards, in 1627 alone, there are 16 occurrences of non. Between 1626 and 1639, the corpus shows 108 occurrences of bare non, which is 63% of all those collected between 1330 and 1700. Before the complete disappearance of the ancient markers around 1600 to the relatively strong emergence of non in 1626, refutation in dramatic texts is ensured by the reduplicated marker non non. Although the phenomenon of reduplication already existed in Medieval French (Saiz-Sánchez 2020b), it does not appear in this corpus of dramatic texts until 1556. ³⁴ The gap created by the disappearance of non + V and nenny is initially filled by non non, until non takes over in the 1620s and 1630s.

Figure 1. Normalised frequency of non + V, nenny, non, and non non between 1330 and 1700

The above results must be interpreted with caution, as the temporal division of our corpus does not lead to temporal sections with an identical number of texts and words. Figure 1 shows the normalised frequency of nenny, non + V, non and non non over the analysed period.

The results do not significantly change but they are put into perspective. The decline in the use of nenny is gradual until it disappears at the beginning of the 17^th century. We still find some residual uses from 1630 onwards (see footnote 33). The use of non + V is more or less stable until 1610, when it disappears as well. The emergence of non non is rather strong in the middle of the 16^th century, compared to the old markers’ frequency. The presence of non non increases progressively until 1660–1675, when it reaches its peak. Before 1626, the frequency of bare non is very low, and there are no occurrences between 1600 and 1625 indeed. Then, between 1626 and 1629, its frequency increases drastically, and decreases progressively afterwards. Between 1620 and 1700, the evolution in the use of non and non non seems reversed: the reduplicated non tends to be more used over time than the simple non, which progressively decreases. In the same period, nenny and non + V are not in use anymore. As said before, non non assumes the refutative role when nenny and non + V disappear, at the beginning of the 17^th century, until non acquires a predominant position in the 1620s.

None of the evolutionary patterns proposed by Ayres-Bennet and Caron (2016: 344) entirely fits the trend observed in this graph. It would correspond to an evolutionary pattern between the first: “the archaic variant disappears completely (after a long period of gradual obsolescence) and the modern variant predominates” although the old forms may still survive “in legal jargon or in a socio- or geo-lect”; and the second pattern: “the balance between the two variants of a variable suddenly changes in a dramatic way at the expense of the archaic one”. In the case of the negation markers, the archaic variants (nenny and non + V) disappear progressively, whereas the new variants (bare non and reduplicated non non) emerge strongly and they are then stabilised in different proportions.

5. CONCLUSIONS

Two conclusions may be drawn from this study. Firstly, the quantitative analysis confirms that the change in the use of French disagreement markers occurs between 1600 and 1620. Although we have no data for 1610–1620, in the decade 1620–1630, the use of the new markers non and non non is well established. The gradual decline of the older forms (nenny and non + V) is precipitated at the end of the 16^th century and leads first to the sudden appearance of reduplicated non non in the middle of the 16^th century. Then, non in its absolute use takes over from 1626 onwards, co-existing in different proportions with its reduplicated version. The 1620s seem to represent as well a turning point for the expression of refutation.

The second conclusion concerns the relevance of keeping the “pre-classical” period in the history of French. It is certainly not a period of stability, but it is a time when old usages disappear at a more accelerated pace and new usages begin to be established, namely the appearance of non non around 1550 and the greater frequency of use of non from 1626, and the subsequent generalisation of both markers afterwards. The notion of chronolect would avoid setting a boundary in this period of constant change. However, the possibility of delimiting the evolution of French temporally using accurate linguistic criteria is still methodologically convenient.

The analysis of the evolution of pragmatic elements coincides with the trends perceived by Ayres-Bennett and Caron (2016) and Amatuzzi et al. (2020) for morphological and syntactic elements. Thus, both findings of this study support the possibility of advancing the final boundary of Pre-Classical French, at least, to the decade 1620–1630, which has been proposed in the former works. This research should also be completed with data from other textual genres in order to confirm the finding for dramatic texts.