1. Introduction
The debate over whether a certain linguistic variety constitutes an autonomous language or a dialect of another variety is typically of little interest to linguists.Footnote 1 No matter whether the variety is spoken by millions or by just a few individuals, what matters to us is that the system analyzed is coherent and that the forms it contains can be argued to be generated by the same mental grammar. Such a question may, however, have far-reaching consequences for endangered European languages that are typologically related to the national language, as only those languages that are “different from the official language(s) of that State” can be recognized and protected under the Charter for Regional and Minority Languages (Council of Europe 1992). Thus, while varieties like Breton, Basque, and Occitan differ sufficiently from French, Spanish or Italian to unequivocally qualify for official recognition and support, regional varieties whose language-versus-dialect status is contentious do not benefit from the same protections. For instance, the Gallo-Romance varieties spoken in Northern France (e.g., Norman, Picard), although now recognized as “languages of France” under Article 75–1 of the 2008 French Constitution, continue to be perceived by many in the greater public as ‘bad’, ‘corrupt’, or, more neutrally, regional varieties of the national language, Continental (or Hexagonal) French (Éloy Reference Éloy and Manzano1997a). Such a perception has contributed to the stigmatization and lack of transmission of these varieties, to their continued exclusion from official school curricula as allowed, for example, by the Deixonne law and the more recent Lang initiative (Éloy Reference Éloy and Manzano1997a), and more generally, to the refusal to grant them official recognition and protection at the national and European levels. In these situations, comparative sociolinguistic research, through its careful examination of variation patterns that focus on both the distribution of variants and the linguistic conditioning behind variant selection, can be of service to language policy makers. Specifically, it can serve as a tool for assessing whether the linguistic distance between two closely related varieties may be sufficient to call them “different languages” such that they can be recognized and protected under the European Charter for Regional and Minority Languages.
Research on endangered languages faces several challenges with respect to data collection and analysis. For instance, it is often difficult to find speakers, especially non-militants (i.e., individuals who are not actively involved in promoting minority language and culture), who are willing to be interviewed, due in part to considerable linguistic insecurity on the part of some community members. In addition, differences in levels of competence and the great diversity that now characterizes the acquisition and usage of these languages have the potential to substantially change the structure of the grammar itself. As a result, limited data often present a significant challenge for quantitative analysis, as they may not provide sufficient token counts for the typical methods of variationist sociolinguistics. Faced with these limits, researchers must find ways to make the most of existing data.
In this article, we focus on Picard and French, two modern reflections of the Oïl dialects spoken in Northern France, to assess the degree to which they share a single grammar. Specifically, we argue that only through an examination of multiple dependent variables which compares rates of (co-)occurrence of Picard-like and French-like variants and, most importantly, linguistic conditioning across the two varieties can a reliable evaluation of morphosyntactic distance between Picard and French be achieved.
2. Disentangling two linguistic varieties
Closely related linguistic varieties often share such large portions of their lexical and grammatical inventories that it is notoriously difficult to draw boundaries between them. In such cases, it is essential to examine multiple linguistic features in order to reach any conclusion. Indeed, although one could argue for differences or similarities between two varieties based on a single dependent variable, such an assessment would fail to recognize that languages are more than an eclectic set of features; instead, variables are part of a cohesive system. Thus, only cumulatively can the various linguistic variables help us disentangle linguistic varieties.
The examination of multiple variables within a variationist sociolinguistic framework has been central to breakthroughs for African American Vernacular English (AAVE), a variety previously misunderstood as a ‘degenerate’ dialect of English. In his early work on AAVE, Labov (Reference Labov1972) analyzed three variables – t/d-deletion, verbal –s, and be deletion – and provided evidence that the variation that characterizes each one is systematic and that each variable contributes to defining AAVE in its own way. A more recent example of this approach is found in Thorburn's (Reference Thorburn2014) analysis of the emerging English variety spoken in Nain, an Inuit community in Labrador, Canada. She examined three variables that include a variant strongly associated with Newfoundland English and found “these salient-in-Newfoundland variants behaving differently in Nain Inuit English” (Thorburn Reference Thorburn2014: 265). She also noted little covariation among the different variables, making it difficult to use one variant's rate to predict the prevalence of another variant belonging to a different variable.
In the European context, the distance between regional minority languages and their respective national languages has been the object of recent work by linguists interested in linguistic convergence (e.g., Morris Reference Morris2013, Nance and Stuart-Smith Reference Nance and Stuart-Smith2013). As more studies analyze variables that can help assess the distance between two genetically-related European varieties, the gap in knowledge slowly narrows. For instance, van Bezooijen (Reference van Bezooijen, Stanford and Preston2009) examined /r/ variation in Dutch and Frisian and showed that the more recent Dutch variants have only been marginally adopted by urban Frisian speakers. In his dissertation on Francoprovençal, Kasstan (Reference Kasstan2015: 316–317) reported that convergence with ‘Standard’ French and the developing regional norm affects variables differently, with the variable (l) showing only incipient convergence toward French, the variable (a) exhibiting some convergence toward the regional norm for Occitan, and plural marking providing evidence of the adoption of new forms from neighbouring Francoprovençal varieties.
Given their genetic proximity, Picard and French are superficially similar in many aspects. Indeed, Picard shares a large portion of its lexicon with French and has a phonological system that greatly overlaps with that of its sister language (Dow Reference Dow2015). Furthermore, it is argued by many linguists (e.g., Carton Reference Carton, Holtus, Metzeltin and Schmitt1990, Pooley Reference Pooley1996, Éloy Reference Éloy1997b, Hornsby Reference Hornsby2006) that its morphosyntax differs minimally from that of colloquial French. This is not to say, however, that the two varieties do not differ in significant ways. For example, lexical items such as rbéyer ‘to watch’, ramintuverie ‘memory’, and maristér ‘primary school teacher’ are obscure to a French speaker who has not been exposed to Picard. Phonological characteristics such as the retention of Germanic /w/ in words like wépe ‘wasp’ (cf. French guêpe) and a different distribution of /ʃ/ and /k/ that produces cacher where French has chasser ‘to hunt’ result in many Picard-French cognates with different phonological representations. The variety's verbal morphology contains many distinct morphemes (e.g., subjunctive –che, 1pl imperfect/conditional –oémes, and overt 3pl /tt/), as can be seen in (1). Finally, differences can indeed be found even in the morphosyntax, with Picard's use of two secondary negations, point and mie (Burnett and Auger Reference Burnett and Auger2016), where French possesses a single one, pas.
-
(1)
-
a. Quoé qu'tu veux qué j'té diche ? (Lettes 18)
What that-you.sg want that I-you say.sbjv
‘What do you want me to tell you?’
-
b. Oz éroémes la djerre. (Lettes 36)
we have.cond.1pl the war
‘We would have war.’
-
c. I creut'té toute. (Lettes 65)
3.nom believe.3pl all
‘They believe everything.’
-
While the characteristics above make it clear that Picard is different from French, extensive and prolonged contact between the two varieties, the severe degree of endangerment that characterizes Picard, and the bilingualism of its speakers have undeniably affected Picard, especially in its spoken form. For example, in her analysis of the oral and written usage of four speakers, Auger (Reference Auger, Joseph, DeStefano, Jacobs and Lehiste2003) revealed that while French-like pas ‘not’ is virtually absent from the texts analyzed, it is used relatively frequently in speech; it is even the dominant secondary negation in the speech of a speaker who is not actively involved in promoting Picard and writing in it. If convergence of this sort is attested in forms that differ in the two varieties, we suspect that structures that are shared may be particularly susceptible to convergence.
In this article, we examine two morphosyntactic structures that Picard shares with French, namely future temporal reference (FTR) and auxiliary alternation (AUX), in order to determine how deep the similarities between the two varieties are. Specifically, we uncover empirical linguistic support for the inclusion of Picard in the list of regional languages recognized by the Direction générale à la langue française et aux langues de France. We argue, as Pooley (Reference Pooley1996: 76–77) did for Chtimi, the variety of Picard spoken in the Nord and Pas-de-Calais départements, that an empirical evaluation of the morphosyntactic distance between Picard and French must consider multiple dependent variables, comparing rates of occurrence of Picard-like and French-like variants and, most importantly, the linguistic factors that condition variation across the two varieties. Thus, we demonstrate that comparative variationist sociolinguistics can play a key role in language planning by helping to determine whether regional varieties closely related to their respective national language are mere dialects of that language or represent different languages eligible for recognition under the European Charter for Regional and Minority Languages.
3. Morphosyntactic distance between Picard and French: the case of Vimeu
Our recent research seeks to assess the degree of structural convergence between French and Picard by analyzing data extracted from sociolinguistic interviews with Picard–French bilinguals and French monolinguals from the Vimeu area, located in Picardie, France. To do so, we examine Vimeu Picard and French oral data from four bilingual men, and Vimeu French oral data from a control group of four monolingual men. The exclusion of women from the corpus, which may seem like an unusual methodological choice, stems from the gender imbalance in the number of regional language speakers and in daily use of the language (Pooley Reference Pooley2003); this makes it difficult to create a balanced sample of female Picard speakers. The Picard data were collected by the first author in the late 1990s; the French data were collected by the second author in the mid–2000s.Footnote 2 These contemporary oral data are contrasted with written data, including an older newspaper column (1946–1966) by G. Vasseur, a key figure in Vimeu Picard.
Table 1: Subjects' demographic information (adapted from Villeneuve and Auger Reference Villeneuve and Auger2013: 119)
In a previous study (Villeneuve and Auger Reference Villeneuve and Auger2013), we analyzed the (co-)occurrence of subject doubling and the preverbal negative clitic ne, two variable features of both Picard and French, in contemporary spoken data extracted from these interviews. Subject doubling refers to the co-occurrence of a subject pronominal clitic with a lexical or strong pronoun subject, as seen in (2a) (compare with non-doubling in (2c)). Negative ne absence, arguably the best studied morphosyntactic variable in French, refers to the variable realization of the preverbal negative clitic ne; this variation is illustrated for Picard in (2a–b) and for French in (2c-d).
-
(2)
-
a. Chés lapins i n’vont poé être gras. (Gérard D., Picard, 69)
the rabbits 3.nom neg-go not be fat
‘The rabbits are not going to be fat.’
-
b. Chés gins i [Ø] cangeoait't point gramint d'poéyi. (Joseph L., Picard, 105)
the people 3.nom changed.3pl not much of-country
‘The people did not move places much.’
-
c. Les volumes [Ø] ne seront pas assez importants. (Joël T., French, 329)
the volumes neg will-be not enough important
‘The production volumes will not be large enough.’
-
d. La première [Ø] [Ø] aime pas beaucoup. (Thomas S., French, 559)
the first like not much
‘The first one does not like [it] much.’
-
When examining subject doubling and negative ne separately, we found that bilinguals, as a group, displayed highly significant differences (p ≤ .001) between the two varieties for both variables. Three of the four bilinguals used subject doubling differently in the two varieties: doubling was (near-)categorical (92%) in Picard, but much lower in French (25%).
The most interesting results, however, emerged when we examined the two dependent variables in conjunction, that is, how subject doubling co-occurs with the presence of ne. We found asymmetries in the distribution of the four possible combinations—from a doubled subject with negative ne (see (2a)) to a lack of both doubling and ne (see (2d))—in the French and Picard speech of bilinguals. In French, although all four combinations are attested, a doubled subject with ne, the least expected pattern in French (Massot Reference Massot2010), is indeed rare. This very pattern, the most frequent in Picard, highlights the contrast between the two varieties.
We now turn to the two variables that are the focus of this paper: future temporal reference (FTR) and auxiliary alternation (AUX).
3.1 Future temporal reference
Like all varieties of French studied to date, contemporary spoken Picard can mark FTR on a verb by adding the appropriate agreement markers onto a future tense stem, as in (3a), or by making use of a periphrasis involving aller ‘go’ followed by the infinitive, as in (3b). These are commonly referred to as the inflected or synthetic future (henceforth IF) and the periphrastic future (henceforth PF), respectively.
-
(3)
-
a. Espérons qu'o fra biétôt la paix. (Lettes 120)
Hope.1pl that-we make-fut soon the peace
‘Let's hope that we will soon make peace.’
-
b. O va meume li foaire eune féte au mouos d'moai. (Lettes 132)
One go even him make a party at month of-May
‘We are even going to throw him a party in May.’
-
While grammarians have commented on the periphrastic form in French for centuries (Poplack and Dion Reference Poplack and Dion2009), Champion (Reference Champion1978: 51) found that for other Gallo-Romance varieties (e.g., Picard, Franco-Provençal), the PF “is mentioned just in passing, if at all, and it apparently plays no important role in the verbal system”.Footnote 3 According to Wilmet (Reference Wilmet1970: 191), the competition between the PF and the IF accelerated in the Middle French period, when Francien was further differentiating itself from other Oïl varieties. Compared to most French varieties documented to date (Emirkanian and Sankoff Reference Emirkanian, Sankoff, Lemieux and Cedergren1985; Poplack and Turpin Reference Poplack and Turpin1999; Blondeau Reference Blondeau2006; Grimm and Nadasdi Reference Grimm and Nadasdi2011; Roberts Reference Roberts2012, Reference Roberts2014; Comeau Reference Comeau2015),Footnote 4 use of IF remains high in traditional Picard. Indeed, the only evidence of PF noted by Champion (Reference Champion1978) in Picard-speaking France (i.e., Chaurand Reference Chaurand1968: 205) comes from Aisne, a département with more intense Picard-to-French shift. Our own survey of Picard grammars and conversation guides—Haigneré (Reference Haigneré1901), Ledieu (Reference Ledieu1909), Debrie (Reference Debrie1983), Vasseur (Reference Vasseur1996), Dawson (Reference Dawson2003)—appears to confirm the somewhat innovative nature of PF in Picard: we failed to find a single mention of the aller + infinitive PF in these sources, despite mentions of a reflexive PF form (i.e., s'in aller + infinitive) in the literature (Debrie Reference Debrie1983).Footnote 5 However, a quick diachronic survey of the large body of Picard written data from the Vimeu area, mostly by Ch'Lanchron authors and members of the Picardisants du Vimeu et du Ponthieu groups, revealed that the PF is attested in written Picard.
In order to measure FTR usage in mid-20th century Picard, we extracted all tokens from G. Vasseur's Lettes à min cousin Polyte, a weekly newspaper chronicle published between 1938 and 1971. To obtain a reliable diachronic data set, we used a ten–month sample from each of 1946, 1956 and 1966. For this assessment of written Picard, and for the analysis of contemporary French and Picard data we present in section 3.1, we defined the variable context as all tokens that unambiguously express that the event will occur after the time of speech. As is customary in variationist studies of FTR, we excluded potential tokens where an IF or PF form has an imperative function (see (4a), from Wagner and Sankoff Reference Wagner and Sankoff2011: 283), or denotes spatial movement or a habitual action (see (4b–c), from Poplack and Turpin Reference Poplack and Turpin1999: 144).
-
(4)
-
a. Tu vas aller t'habiller puis tu vas sortir avec moi.
you go to-go 2sg.refl-dress then you go to-go-out with me.
‘Go get dressed, then you're going out with me.’
-
b. On va chercher Adelaïde pour jouer au piano.
we go search-for Adelaïde in-order-to play to-the piano
‘Let's go find Adelaïde and play the piano.’
-
c. Des fois ils vont changer des affaires parce que ça plaira pas
of-the times they go change of-the things because it please.fut not
au monde.
to-the world
‘Sometimes they change things because people don't like them.’
-
We also examined the linguistic conditioning affecting variant choice in our sample. We coded each token for a series of factors described in the literature as having an effect on this variable in French: temporal distance, sentential polarity, and the presence of a temporal adverbial (Poplack and Turpin Reference Poplack and Turpin1999, Comeau Reference Comeau2015, Grimm Reference Grimm2015). More specifically, proximate future events have been shown to favour the PF in spoken Acadian (King and Nadasdi Reference King and Nadasdi2003, Comeau Reference Comeau2015), Martinique (Roberts Reference Roberts2014) and Picardie French varieties (Villeneuve and Comeau Reference Villeneuve and Comeau2016); this tendency is consistent with grammarians' description of this variant as le futur proche ‘the proximate future’. In other varieties, temporal distance played little or no role; a polarity constraint is in effect in some of them, according to which negative sentences strongly favour the IF (Poplack and Turpin Reference Poplack and Turpin1999, Blondeau Reference Blondeau2006, Wagner and Sankoff Reference Wagner and Sankoff2011, Roberts Reference Roberts2012). The presence of temporal adverbials has also been shown to favour the IF (Emirkanian and Sankoff Reference Emirkanian, Sankoff, Lemieux and Cedergren1985), although this effect was recently described as “minor” (Poplack and Dion Reference Poplack and Dion2009: 573).
In our written Picard corpus (N = 393), the PF represents on average 16.5% of FTR forms (see Table 2). It is clear that the IF is the default marker of futurity in Picard between 1946 and 1966, and this distribution remains somewhat stable over this period (χ2(2) = 3.136; p =.208).
Table 2: Overall distribution of FTR variants in written Vimeu Picard (1946–1966)
A multivariate analysis was performed with Goldvarb (Robinson et al. Reference Robinson, Lawrence and Tagliamonte2001), setting the PF as the application value; results are presented in Table 3. Due to uneven distributions in 1946 and 1956 and to the similar trends found in both years, we combined these data (Period I) and contrast them with 1966 (Period II) in order to improve the predictive strength of the model. The slight increase in input value from Period I (.11) to Period II (.18) indicates that the PF, although still a minority variant in mid–20th century Picard, appeared to be rising slowly. Although the increase is insignificant for 1946–1966, we expect a higher proportion of the PF in our oral Picard data from the late 1990s. Given the high rates of the PF reported in most non-Acadian French varieties, we hypothesize that Picardie French will display high rates of the PF, perhaps even higher than in Picard.
Table 3: Linguistic conditioning of FTR in written Vimeu Picard (1946–1966)
As for linguistic constraints affecting variant choice in written Picard, Table 3 shows that sentential polarity is the only factor group affecting FTR significantly across the 20–year period: negative contexts strongly disfavour the PF. However, the polarity constraint weakens from Period I to Period II, as indicated by the narrowing in range from 52 to 41. Despite not reaching statistical significance by 1966, the effect of temporal distance appears to be increasing. Adverbial specification does not have any significant impact on the choice of FTR variant in this sample of Vasseur's Lettes. Based on these results, we expect an even weaker polarity effect and perhaps a slight effect of temporal distance in our 1990s oral Picard data.
3.2 Auxiliary selection
In French, most verbs are conjugated with avoir ‘have’ in compound tenses. There exist, however, two types of verbs that must be conjugated with être ‘be’ according to prescriptive grammars: a small number of intransitive and state verbs (see Table 4 for a complete list) and all pronominal verbs; see (5) for examples. Actual usage differs considerably from that prescription, as use of avoir has long been reported in colloquial varieties (e.g., Bauche 1920/Reference Bauche1946: 121; Frei 1929/Reference Frei1971: 86, 166; Gadet Reference Gadet1992: 55). Variationist studies such as Sankoff and Thibault (Reference Sankoff and Thibault1977), Russo and Roberts (Reference Russo and Roberts1999), and Stelling (Reference Stelling2008) have revealed that linguistic and social constraints influence the degree to which avoir is attested. To our knowledge, Acadian French is the only French variety that has virtually eliminated être as an auxiliary. King and Nadasdi (Reference King, Nadasdi, Balcom, Beaulieu and Chevalier2001) report categorical use of avoir in Acadian French varieties spoken in Prince Edward Island and in Newfoundland, with the exception of two verbs, naître ‘be born’ and mourir ‘die’, while Roussel (Reference Roussel2016) suggests that the generalization process is less advanced in the ‘Acadian peninsula’ of New Brunswick.
-
(5)
-
a. Le DEPP c’est devenu l'concours d'entrée en sixième. (Joseph L., 517)
the DEPP it-is become the-contest of-entry in sixth
‘The DEPP has become the entrance exam for sixth grade.’
-
b. quand j’suis parti au lycée à Abbeville (Thomas S., 320)
when I-am left to-the high-school at Abbeville
‘when I left for high school in Abbeville’
-
c. J'me suis absenté trois semaines. (Joël T., 664)
I-1sg-refl am absented three weeks
‘I went away for three weeks.’
-
Table 4: Verbs conjugated with être in French
According to Vasseur (Reference Vasseur1996: 79–80), avoir is the only auxiliary used in the conjugation of all Picard verbs; the only exception is mourir ‘die’, which is conjugated with éte (Vasseur Reference Vasseur1996: 52).Footnote 6 Given that, with the exception of Acadian varieties, use of être as an auxiliary is frequent in French, this would appear to be a clear morphosyntactic difference between Picard and French. However, once again, actual usage reveals considerable variation, as shown in (6), both uttered by the same speaker in the context of a radio interview.
-
(6)
-
a. J’ai vnu au monne à l’Étoéle. (J. Leclercq, PVADP)
I-have come to-the world at L’Étoéle
‘I was born in L’Étoéle.’
-
b. Éj sus vnu au monne à l’Étoéle. (J. Leclercq, PVADP)
I am come to-the world at L’Étoéle
‘I was born in L’Étoéle.’
-
Since studies of non-Acadian French varieties (e.g., Sankoff and Thibault Reference Sankoff and Thibault1977, Renaud and Villeneuve Reference Renaud and Villeneuve2008) have found that auxiliary variation is limited to non-pronominal verbs and that our French data contain no instances of avoir with pronominal verbs (N = 67 for bilinguals; N = 23 for monolinguals), we focus our comparison of Picard and French on non-pronominal verbs.
Our quantitative analysis excludes examples in which no variation is possible between the two auxiliaries. Thus, verbs of manner of movement (e.g., marcher ‘walk’, courir ‘run’) are excluded because they are invariably conjugated with avoir, as seen in (7a). Verb meaning is another factor that affects choice of auxiliary. For example, even though partir can be conjugated with être when it means ‘depart’, its use to mark inchoative aspect requires avoir; see (7b). Likewise, when an intransitive verb is used transitively, avoir must be used, as seen in (7c). Conversely, the passive voice requires être as an auxiliary, as in (7d). Finally, when the past participle is interpreted as an adjective, the construction involves copula être and use of avoir is ruled out, as seen in (8a), where the speaker is referring to the absence of the English woman rather than to her act of leaving. In some cases, however, both auxiliary and copula interpretations are possible. Ambiguous cases such as (8b) are excluded from our analysis in order to avoid artificially inflating the number of être uses. While all the examples below are from Picard, the exclusions discussed above characterize both languages.
-
(7)
-
a. Quante j'ai seu o, j’ai couru à Paris pour vir no ministe. (Lettes 200)
when I-have known it, I-have run to Paris to see our minister
‘When I heard this, I rushed to Paris to see our minister.’
-
b. À trouos heures d’érmontèe, ch'vint il a parti à souffleu
at three hours of-afternoon, the-wind he has left to blow
coér pu fort. (Bos blond 18)
still more strong
‘At 3 o'clock in the afternoon, the wind started to blow even harder.’
-
c. Il a montè eune boutique, él boutique éd Mossieu Pérmintieu. (Lettes 519)
he has raised a shop, the shop of Mr. Parmentier
‘He started a shop, Mr. Parmentier's shop.’
-
d. Si leu constitution a n’est po coére votée au mouos d'octobe ? (Lettes 154)
if their constitution she neg-is not still voted at.the month of-October
‘If their constitution is still not adopted in October?’
-
-
(8)
-
a. Oui, min fiu, no pieute Inglaise al est partie ! (Chl'autocar 102)
Yes, my son, our little English-woman she is left
‘Yes, my son, our little English lady has left.’
-
b. Quante tu rchuvaros m'lette, chés piots i sront rpartis à l’école. (Lettes 210)
When you.sg receive-fut my-letter, the little 3.nom be-fut left-again to the-school
‘When you receive this letter, the kids will be back in school/have left for school.’
-
Based on the existing literature, we expect Picard and French to differ with respect to both of these variables. Specifically, we expect rates of use of the PF to be much higher in French than in Picard, while use of the avoér auxiliary will be much lower in French than in Picard.
4. FTR and AUX in Vimeu Picard and French
Table 5 presents the overall frequencies of FTR and AUX variants in French and Picard. Contrary to our hypothesis, as a group, bilinguals use the PF to the same extent in Picard and in French, and at a rate comparable to that of French monolinguals. The difference between Picard and French is, however, much clearer for AUX, where bilinguals – in the case of intransitive and state verbs – avoid the use of avoir categorically in French, but use both variants almost equally in Picard, as we will discuss in section 4.2.
Table 5: (FTR) and (AUX) in Vimeu French and Picard
4.1 Future temporal reference
Looking first at the FTR variable, we see great similarities in French between the two speaker groups, and between the French and Picard output of bilinguals. In fact, the distribution of French FTR variants in monolinguals is not significantly different from that found in bilinguals (χ2(1) = 1.081; p = .298): both speaker groups preferred the periphrastic form six to seven times out of ten. However, the small number of tokens for several speakers prevents us from making generalizations about individual variation in French at this point.
There is no significant difference between the French and Picard output of bilinguals either, whether we consider the four speakers as a group (χ2(1) = 0.045; p = .832) or individually (Fisher exact p = .707Footnote 7 for Gérard D.; p = .155 for Joël T.; p = .123 for Thomas S.). What might explain these similarities? Do bilingual speakers associate the PF with colloquial French, leading them to make greater use of this form in Picard, or is the variation between the PF and the IF below the level of consciousness? The latter scenario, which we view as more plausible given the lack of stigmatization of either FTR variant, may have gradually led to the greater degree of similarity between the two varieties through their prolonged contact in Vimeu. But, as we found to be the case with respect to subject doubling (Villeneuve and Auger Reference Villeneuve and Auger2013), it is also possible that the distributional similarity could be hiding distinctions in the linguistic conditioning of FTR in French and in Picard.
As we saw in our analysis of Vasseur's Lettes in section 2, sentential polarity is the only linguistic predictor of FTR variant choice in mid–20th century written Picard. Previous studies have shown that the linguistic conditioning in French is more complex: regional varieties appear to fall along a temporal distance–polarity divide, with other factors (e.g., temporal adverbials, contingency on a si clause) having a minor effect on the variable. Therefore, we submitted the French and Picard data to multivariate analyses which tested the effect of three factors: sentential polarity, temporal distance and the presence of a temporal adverbial. We ran a separate analysis for each of three data sets to assess structural similarities between groups in French (bilinguals versus French monolinguals) and between varieties (French versus Picard) for bilinguals. Results are presented in Table 6. The three input values (.74, .67, .63) reflect the fact that differences in relative frequency of the two variants are minor within and across varieties, as we just saw. However, differences are found in the linguistic factors conditioning FTR in contemporary oral Picard and in French. While oral Picard shows the same polarity effect present in Vasseur's mid–20th century Picard, we find an additional constraint in the more recent oral data: the presence of a temporal adverbial disfavours the PF. The effect of polarity in French, although in the expected direction – negatives lower the probability of a PF – and more noticeable in bilinguals' speech (.53 versus .25) than in monolinguals' (.51 versus .46), is not statistically significant. The role of adverbials, and that of temporal distance, fail to reach significance as well.
Table 6: Linguistic conditioning of FTR in contemporary oral French and Picard
Given the polarity effect, significant in Picard and apparent in bilinguals' French, and the association of the IF with overt ne in Vimeu French (Villeneuve and Comeau Reference Villeneuve and Comeau2016), we teased apart the effect of polarity and that of ne by examining negative contexts further, factoring in the variable presence of ne. Figure 1 shows the distribution of four co-occurrence patterns for a FTR variant and ne in negative contexts. When ne is absent, the PF is used more than half the time across all three data sets, and the effect of polarity seems to be exacerbated by its interaction with ne in bilinguals: the IF is categorical when ne is present, in French and in Picard.Footnote 8 Since both the IF and ne are associated with normative French and with normative Picard, it is difficult to disentangle the role of style in each variety from that of Picard-to-French (or French-to-Picard) linguistic transfer. The co-occurrence of the IF and ne appears more likely for bilinguals, but this pattern does not differ in the two varieties (Fisher exact p = .846).
Figure 1: FTR and negative ne in Vimeu French and Picard
In short, the comparison of linguistic constraints in Vasseur's writings and our oral data uncovered three pieces of evidence pointing to structural differences between the two varieties: the enduring polarity constraint in Picard along with its absence from French, the emergence of an adverbial effect in recent oral Picard but not in French, and the fact that the only linguistic factor found significant for a larger sample of Vimeu French, temporal distance (Villeneuve and Comeau Reference Villeneuve and Comeau2016), fails to materialize in recent Picard. As for the apparent convergence in FTR and ne co-occurrence patterns, it is accounted for by the fact that neither variable is salient in colloquial French, and that the IF and overt ne share the same sociolinguistic value in Picard and in French.
Having established a certain degree of similarity in the distribution of FTR variants as well as subtle differences in linguistic conditioning, we now turn to the results for our second variable, auxiliary alternation, which reveal a much greater distance between Picard and French.
4.2 Auxiliary alternation
Our French interviews contain 154 tokens of non-pronominal verbs that can be conjugated with either avoir or être. As we saw in Table 5, use of avoir is very infrequent in these data. In fact, it is completely absent from our bilinguals' 87 tokens. The only three tokens of avoir (4.5%, N = 67) come from monolingual speakers and are used with verbs—rester ‘stay’, descendre ‘go down’ and sortir ‘go out’—which have been linked to higher use of avoir in Montréal French (Sankoff and Thibault Reference Sankoff and Thibault1977).
The data summarized in Table 5 show that use of avoér by bilinguals is significantly higher in Picard than in French (χ2(1) = 60.337; p = 0), with an overall rate of 52.8%. However, a closer look at the individual rates reveals an important difference: Joël T. and Gérard D. have much lower rates than Joseph L. and Thomas S. For the speakers with the lower rates, it should be mentioned that (i) those rates are higher than any individual rate observed in French and (ii) neither speaker uses avoir in French. One possible explanation for this difference between the two pairs of speakers can be found in the fact that the former do not write, or write very little, in Picard, while the latter have both published numerous texts in this language. As Auger (Reference Auger, Joseph, DeStefano, Jacobs and Lehiste2003) showed, when Picardisants may choose between a variant that is shared with French and a variant that is perceived as Picard,Footnote 9 use of the latter is highest in written production, intermediate in the oral production of authors, and lowest in speakers who do not write in Picard. Such a pattern is interpreted as evidence that authors are more aware of what distinguishes the two varieties and have a desire, possibly unconscious, to distinguish the two closely related varieties.Footnote 10 Support for such an awareness can be found in (9), where Jean Leclercq first uses éte and then corrects himself by using avoér. We can also draw a parallel with the results reported in Villeneuve and Auger (Reference Villeneuve and Auger2013): Joël T. and Gérard D. had the highest rates of ne absence in Picard. Unlike many features which are shared by Picard and colloquial French, ne absence is typical of the latter but not of the former. And, whereas three speakers used subject doubling at rates much lower in French than in Picard, Gérard D. had a high rate of subject doubling in French, 65%, which suggests that his French is affected by Picard for this structure.
-
(9) Gaston Vasseur, il est vnu – il a vnu étou écouter des canchons. (PVADP)
Gaston Vasseur 3.nom is come – 3.nom has come too to.listen of.the songs
‘Gaston Vasseur came—came, too, to listen to songs.’
4.2.1 Auxiliary selection in non-pronominal and pronominal verbs
As we mentioned above, pronominal verbs in our French data are categorically conjugated with être. Given that our Picard data contain many tokens of avoér with pronominal verbs, we now turn to this variety and compare auxiliary usage in non-pronominal and pronominal verbs in order to determine whether contact with French is the only source of this variation or whether language-internal conditioning may explain some of it.
As we see in Table 7, use of avoér is almost as high with pronominal as with non-pronominal verbs, with a frequency of 46.4% compared to 52.8% (χ2(1) = 0.325; p = .569). Unfortunately, Picard token counts for Joseph L. and Gérard D. are too low to determine whether their usage with pronominal verbs mirrors their rates with non-pronominal verbs. For Joël T. and Thomas S., we do observe very similar rates for both verb types (Fisher exact p = 1.000 for both speakers), which suggests that AUX operates similarly with the two types of verbs.
Table 7: (AUX) in spoken Picard: pronominal vs. non-pronominal verbs
A closer look at auxiliary choice reveals that it varies considerably across persons, especially for non-pronominal verbs where these differences are significant (Fisher exact p = .0005), as shown in Table 8. We see that whereas 1sg favours avoér and auxiliaries are equally frequent in 3sg, éte is strongly favoured in 3pl. Despite small numbers of tokens (making it difficult to reach statistical significance for pronominals), we would like to argue that formal similarity between éte and avoér is one important reason for the variation observed and the high rate of éte in 3pl.
Table 8: Non-pronominal vs. pronominal verbs and subject person (oral Picard)
Table 9: Present tense of avoér and éte (Vasseur Reference Vasseur1996: 35–36)
Table 9 shows that in the present tense, there is no formal similarity between the two auxiliaries in 1sg, 1pl, and 2pl. However, in 2sg and 3sg, each auxiliary is realized by a single vowel. Though these vowels are quite different, the fact that they typically occur in unstressed positions may create confusion between the two forms. In the case of 3pl, the presence of an initial /s/ distinguishes éte from avoér. If we add to this the fact that one of the two 3pl reflexive pronouns is /s/, we suggest that the near-homophony between i s'ont [isɔ̃] and i s'sont [issɔ̃] significantly favours the use of éte in this context.Footnote 12 Our five tokens of éte in 3pl in pronominal verbs do involve éte in the present tense and the clitic s'; see (10) for an example. We offer further evidence of the association of reflexive pronoun s' with French in section 4.2.2.
-
(10) I s'sont rtournés, pis iz noz ont rbéyé. (Thomas S., 1031)
3.nom 3.refl-are turned.around and 3pl.nom ushave.3pl looked-at
‘They turned around, and they looked at us.’
While the difference between a single and geminate [s] can easily cause perceptual confusion, an explanation remains to be provided for the non-pronominal cases. Indeed, given that subject doubling is near-categorical in Picard, the auxiliaries are virtually always preceded by il ‘3.nom’, thus producing [ilɔ̃] for avoér and [isɔ̃]Footnote 13 for éte. However, il is not the only 3pl nominative pronoun in Vimeu Picard. In addition to the il form documented in Vasseur (Reference Vasseur1996: 32), there is frequent use of an iz variant by some speakers. In Table 10, we report pronoun usage for our four Picard speakers, as well as for some texts published through the 20th century and in the 21st century. As we can see, there is considerable variation across speakers/authors, with some being categorical il ont users and others categorical iz ont users, as well as some intraspeaker variation. Given that the only difference between iz ont [izɔ̃] and i sont [isɔ̃] is the voicing of the fricative and that il ont users interact with iz ont users, we submit that this variation gives rise to more uncertainty considering auxiliary choice with 3pl verbs.
Table 10: il ont vs. iz/is ont in oral and written data
4.2.2 Co-occurrences
One final piece of evidence in support of our hypothesis comes from an analysis of AUX based on tense and person in G. Vasseur's Lettes. Our findings for all six persons and the three compound tenses that are well represented in this text reveal that the use of éte is marginal with pronominal verbs in Vasseur's grammar (3.6% éte; N = 335) and that the only context in which it is the dominant variant is the 3pl passé composé when the reflexive pronoun is s'(88.9% éte; N = 9). This contrasts with all other contexts where the use of avoér is near-categorical, including the 55 tokens in which the 3pl reflexive pronoun is leuz; see (11). This sharp contrast between s' and éte, on the one hand, and leuz and avoér on the other (Fisher exact p ≤ .00001), is accounted for by the fact that the first pair is associated with French, while the latter is associated with Picard. Indeed, while s' functions as a 3pl reflexive pronoun in both languages, leur is a 3pl dative pronoun in French. Only in Picard is leuz ambiguous between its dative and reflexive uses (Vasseur 1996: 75). Furthermore, as we saw above, Vasseur's (Reference Vasseur1996) grammar presents avoér as the only auxiliary used in Picard.
Given that the contrast between leuz ont [løzɔ̃] and leu sont [løsɔ̃] also relies on the voicing of the fricative, we would expect to observe the same type of variation as was described above. However, we do not. We hypothesize that the strong association of reflexive leuz with Picard is sufficient to prevent the possible encroachment of the French-like auxiliary in clauses in which leuz is used. Example (12), which features both types of pronominal use in the same sentence, supports an association between the French-like s' pronoun and éte auxiliary that contrasts with the Picard character associated with reflexive leuz and the avoér auxiliary.
-
(11) Deux heures durant, i leuz ont proumenè. (Lettes 1339)
Two hours during 3.nom 3pl.refl have walked
‘For two hours, they walked around.’
-
(12) I leuz ont chtè l'un dsu l'eute et pi i s'sont
3.nom 3pl.refl have jumped the-one on the-other and 3.nom 3.refl-are
tuès tortous. (Lettes 28)
killed all
‘They attacked each other and they killed each other off.’
Before concluding, we turn to another co-occurrence pattern: that of auxiliary choice and negative ne. Given that our French data contain only three tokens of auxiliary avoir and that the only one within a negative clause is ambiguous with respect to the presence of ne due to the presence of the pronoun on in a prevocalic context, we cannot contrast our Picard data with our French data. But an analysis of the nine tokens of compound tenses in negative clauses in our oral Picard data reveals a complementary distribution that is quite unexpected in French:Footnote 14 all five tokens of avoér occur in the presence of an overt n’, while the four tokens of éte occur in clauses in which n’ is absent; see (13). As we reported for reflexive leuz and the avoér auxiliary, we observe a strong tendency for forms considered to be Picard to co-occur. Although we acknowledge that ne could also be constrained by phonology,Footnote 15 we interpret this complementary distribution as one more piece of evidence that forms perceived as belonging to the same grammar are more likely to co-occur than those associated with different grammars.
-
(13)
-
a. i sont point vnus s'batte (Thomas S., 281)
3.nom are not come 3.refl-fight
‘they didn't come to fight’
-
b. i nn'a granmint qu'i n'ont point rparti (Thomas S., 327)
3.nom of.it-has a.lot that-3.nom neg-have not again.left
‘there are many who haven't left’
-
5. Conclusion
In the context of the ongoing debate over the status of Northern France's obsolescent varieties, our Picardie data provide empirical evidence for the fact that bilinguals maintain a mental grammar in the minority language that is distinct from that of the national language. The two variables analyzed in this paper, AUX and FTR, look quite different based on the overall distribution of the variants in the two varieties, confirming a pattern previously observed for two other variables in these varieties (Villeneuve and Auger Reference Villeneuve and Auger2013). The difference between French and Picard is quite apparent in the case of AUX: while use of avoir with intransitive and state verbs is truly marginal in French, it represents roughly half of the tokens in Picard. Additionally, it has been shown that co-occurrence of avoér with overt negative ne in Picard differs from what is expected in French, that is, avoir and ne absence. In the case of FTR, an increased frequency of the periphrastic future compared to what is observed in older written data, and frequencies that are not statistically different in the French and the Picard of bilingual speakers suggest considerable convergence of Picard toward French. However, a variationist analysis of the effect of polarity, temporal distance, and overt presence of a temporal adverbial reveals quite different systems. Not only does oral Picard share with older written Picard the fact that negative polarity favours use of the IF, but in the latter data set, adverbials were also found to play a significant role in variant selection. Neither linguistic factor played a significant role in our French data.
As we mentioned in our introduction, analyses of endangered languages are often complicated by somewhat small corpora and token counts that may not lend themselves to the sort of quantitative analyses that are customary in variationist sociolinguistics. Without such analyses, it may prove difficult to provide solid evidence concerning the extent to which the variable patterns of two varieties differ. Furthermore, analyses based on a single variable may reach conflicting conclusions concerning the extent to which an endangered language maintains a grammar distinct from that of its close parent, as superficial convergence can be observed with some variables but not others. The findings presented in this article provide a strong argument for investigating multiple variables before reaching any conclusions.
Our results in this study mirror, to a great extent, those in Villeneuve and Auger (Reference Villeneuve and Auger2013), where we reported that subject doubling was used with different rates in the two varieties but that the rates of ne were not different across varieties. We argued that the co-occurrence pattern between subject doubling and ne clearly distinguished Picard and French. Similarly, the present study reveals superficial convergence in the use of FTR variants, but not in the case of AUX. Furthermore, the co-occurrence patterns for Picard reported in this paper reveal a strong affinity between variants perceived as Picard, which results in patterns quite unlike those of French when the variants in question have different socio-stylistic values in the two varieties. Interestingly, in both studies, the variables most subject to convergence are the least salient to their speakers. Indeed, ne is so infrequent in current colloquial usage that its absence is rarely noticed by speakers. As for the variation between the IF and the PF, although the former undoubtedly maintains a higher degree of prestige, as can be inferred from its increased use by Montréal middle-class speakers as they age (Wagner and Sankoff Reference Wagner and Sankoff2011), both variants are acceptable in written and formal settings, and PF usage appears to receive relatively little attention on the part of teachers, grammarians or language prescriptivists (Poplack and Dion Reference Poplack and Dion2009: 561).
Given the disparity between Vasseur's (Reference Vasseur1996) claim that avoér is the only auxiliary in Picard and the 52.8% frequency reported in our study, the possibility that the high frequency of éte could be due to convergence toward French should be examined. While we cannot rule out this possibility, we have reasons to doubt that auxiliary usage has, in reality, ever been as categorical as suggested by Vasseur's grammar. Unfortunately, oral data from earlier periods are not available to assess whether that was the case and determine whether the variation observed in the contemporary data must be attributed to intense contact with French. However, the variation documented in his own writing provides one piece of evidence that usage was more variable than claimed by Vasseur himself. Additional examples from two contemporary authors in (14) show that éte was attested in their usage, too. As we can see in Table 11, only in the texts of younger Picardisants from the late–20th century do we find the kind of categorical usage described by Vasseur. We attribute this finding to the ongoing effort to create a literary standard for Picard and the concurrent effort for distancing Picard from French (Auger Reference Auger, Joseph, DeStefano, Jacobs and Lehiste2003). Indeed, favouring use of avoér in contexts in which schoolteachers and prescriptivists have been trying to eradicate its use in French makes it clear that the literary standard for Picard is distinct from the French standard. Given the sharp contrast reported by Auger (Reference Auger, Joseph, DeStefano, Jacobs and Lehiste2003) for the spoken and written usage of three Picardisants and the diachronic decline in use of éte as an auxiliary, it would appear that, like subject doubling, use of the avoér auxiliary may have become a badge of Picard identity.
-
(14)
-
a. Mais i n'n’étoait'té point coère évnus à bout Adville. (Ch'coin 41–42)
but 3.nom neg-of-it-were not yet come to end Abbeville
‘But they hadn't managed to end it [with the Germans] in Abbeville.’
-
b. Alle est v'nue ein vagances pour quiques jours. (Gronnée 57)
she is come in vacation for a.few days
‘She's come for a holiday for a few days.’
-
Table 11: (AUX) in written Picard: diachronic data
The results from the current study raise questions that will need to be addressed in our future research. What accounts for the markedly higher rate of PF and the difference in linguistic conditioning (i.e., the emergence of adverbials as a significant factor) between Vasseur's 1946–1966 written Picard and contemporary oral Picard? Is this linguistic change at the community level, does it reflect a modal distinction between oral and written Picard, or is this a case of idiolectal variation? If it is indeed change, what precipitated it? In collecting the FTR data, we found a clear difference in the frequency of habituals and gnomics between our 1946–1966 Picard corpus, where we only found a handful, and our more recent oral data, where they constitute a larger proportion of aller + infinitive forms, as in contemporary French (Wagner and Sankoff Reference Wagner and Sankoff2011: 279). Although these ‘false futures’ are excluded from the FTR envelope of variation, could the greater frequency of the periphrasis in one context favour its increase in another? To answer these questions, further investigation must involve a larger sample of Vasseur's work over the years but also include a comparison with more recent written Picard. The availability of both oral and written data for some of our bilingual speakers may also help answer some of these questions.
In addition to expanding the study of language variation and change to regional minority languages, which are less commonly investigated in the quantitative sociolinguistic paradigm, this work highlights how the field can benefit from studies set in long-standing language contact areas and how comparative sociolinguistics can inform language policies. In the context of the European Charter for Regional or Minority Languages, which only applies to “different” languages and not “dialects”, the comparative method developed by Poplack provides a strong analytical tool for assessing the extent to which superficially similar forms in closely related varieties are constrained by the same linguistic and social factors. In this study, its use has shown that, in spite of its “seriously endangered” status (Duriez Reference Duriez2009: 9), Picard continues to exist as an autonomous language, and does not share its underlying grammar with the local variety of French. The variation that characterizes it is the product of intricate patterns of linguistic conditioning that can only result from an internalized full-fledged Picard grammar. In other words, this linguistic variety deserves to be recognized by France as a language in its own right and supported under the European Charter for Regional and Minority Languages.
