Article body

1. Introduction

Because the cooccurrence of lexical subjects and coreferential subject clitics is variable in contemporary colloquial French, many linguists are hesitant to analyze this construction as subject doubling and view subject clitics as agreement markers. However, this is not the case for Picard, a Gallo-Romance language closely related to French, as virtually all authors present subject doubling as a categorical phenomenon in this language (e.g., Edmont 1897: 10 for Pas-de-Calais; Ledieu 1909: 42, Hrkal 1910: 262, and Debrie 1974: 18 for Amiénois; Cochet 1933: 36 and Dauby 1979: 43 for the Nord département; and Vasseur 1996: 61 for Vimeu). A few examples of subject doubling are presented in (1), where lexical and pronominal subjects co-occur with subject clitics that share their grammatical features.

This double expression of the subject, along with other properties that will be summarized below, has led me to argue that in such constructions, the noun phrase/strong pronoun functions as the syntactic subject and that the subject clitic is an agreement marker (Auger 2003a, b). While this analysis predicts that subject clitics should occur in all clauses that contain a tensed verb, careful analysis of Picard uncovers a small, though significant, number of clauses in which no subject clitics are present. Interestingly, clitic absence is not random, as we will see below. More interestingly, still, one group of exceptions opens a window into a rather complex system of neuter clitic pronouns that appears to have developed in Picard.

This paper opens with a brief description of subject doubling in Picard. It then describes the neuter subject clitic system, proposes a hypothesis concerning its origin, compares it with neuter pronouns in French and other Gallo-Romance languages, and determines what governs the choice between forms that share the same grammatical meaning and features but occur in complementary distribution. Given that Picard is spoken in a large territory that includes much of northern France and parts of Belgium, I will limit my discussion to the system that characterizes the variety of Picard spoken in Vimeu, in the western part of the Somme département in France.[2] My corpus consists of literary texts in prose written by Picard speakers born in Vimeu or the adjacent town of Abbeville between 1904 and 1959 and published between 1938 and the 21st century; it also includes oral recordings collected during fieldwork in the 1990’s.

2. Subject doubling in Picard

In this section, I briefly review aspects of subject doubling that support an analysis in which the lexical phrase or strong pronoun constitutes the syntactic subject and the pronominal clitic functions as a preverbal agreement marker over an analysis in terms of dislocation.

2.1 Pronominal clitics as agreement markers

Sentences such as those in (1), in which the subject appears to be expressed twice, raise questions concerning their structure. Given the widely accepted view that verbs can assign nominative case to only one subject, the possibility that both the DP/strong pronoun and the clitic are subjects is ruled out. This leaves the analyst with a choice between two possible structures: either the clitic is the syntactic subject and the DP/strong pronoun is a dislocated phrase that is set apart from the core sentence to achieve some pragmatic effect (e.g., emphasis, contrast, introduction of a new topic; cf. Barnes 1985 and Ashby 1988), or the DP/strong pronoun fulfills the subject function and the pronominal clitic has been reanalyzed as a preverbal agreement marker. While the former analysis is typically adopted for standard French (King & Nadasdi 1997: 269), many researchers have argued that the latter analysis better describes colloquial French (e.g., Roberge 1990, Auger 1994, Zribi-Hertz 1994, and Culbertson 2010; but see King & Nadasdi 1997 for discussion of a variety in which subject clitics have not been reanalyzed as agreement markers and de Cat 2005 for a rejection of this analysis for colloquial French). Those who favor the agreement marking analysis base their position on the fact that in these varieties of French, subject clitics occur in all contexts in which a verb is expected to agree with its subject. For instance, in colloquial Québec French, subject clitics cooccur with bare-quantifier subjects, (2a), a type of subject that is incompatible with a dislocated position (Rizzi 1986), they are repeated on each conjunct in a VP-conjunction structure, (2b), and they occur in subject relative clauses, (2c), as well as in inverted constructions, (2d).

The criteria that support an agreement-marking analysis for colloquial French subject clitics provide support for the same analysis in Picard. In this language, all subjects are doubled, including bare quantifiers, as seen in (3b). Subject clitics are present in subject-verb inversion constructions, (4); they are repeated on each verb in a VP-conjunction, (5); and they occur in subject relative clauses, (6). In subject wh-questions, a default third person masculine singular marker is used, as illustrated in (7).

While the subject pronouns of Picard function as agreement markers and we might expect them to have been reanalyzed as lexical affixes, Auger (2003a) argues that they still are clitics. She bases her conclusion on the fact that subject clitics are unaffected by phonological rules that apply word-internally and that their combination with other pronominal clitics and their verbal hosts differs from that observed at word boundaries. We can capture the clitic status and agreement-marking properties of Picard subject clitics by generating them under AgrS; this structure allows them to co-occur with overt subjects and to combine with verbs that may bear overt agreement markers of their own, as shown in (8).

2.2 Apparent exceptions

2.2.1 Subject doubling

While subject doubling is, as we saw above, described as applying categorically in Picard, a careful analysis of written texts that span the 20th century reveals two constructions in which a subject clitic does not cooccur with a lexical subject. The first exception concerns quantified subjects, especially bare quantifiers. For some speakers, most often older ones, subject doubling does not occur with subjects such as parsonne ‘nobody’ and tout le monne ‘everyone’, (9). Given the use of doubled bare quantifiers by younger speakers, (10), we can conclude that the grammaticalization process, which was still incomplete in the middle of the twentieth century, has reached its completion among the latest generation.[3]

The second type of exception, illustrated in (11), involves constructions in which French would use the neuter clitic ce. Evidence that we should analyze such structures as involving a null subject clitic rather than clauses lacking subject doubling is provided by (11b): here, we see a feminine singular subject, eine ruque, cooccurring with a masculine predicate adjective. Given that subjects and predicate adjectives normally agree in number and gender, as seen in (11c), an explanation must be sought for this apparent clash. 

2.2.2 Clauses with no overt subject

As noted previously, if subject clitics are agreement markers, they should occur with every finite verb. However, there are two constructions in which subject expression appears to be completely absent. The first type, illustrated in (12), is parallel to similar sentences in colloquial French in which expletive il can be omitted. The second type, seen in (13), has no parallel in colloquial French; however, quite interestingly, all examples would involve the neuter pronoun, just like the second type of exception to subject doubling described above.

3. Neuter pronouns in Picard 

I propose that in (11a), (11b), and (13), a phonologically null neuter subject clitic is present. As is the case with ce/ça[4] in French, this clitic imposes default masculine singular features on predicate adjectives and verbs. This analysis explains the apparent clash over gender agreement in (11b), as well as the apparent clash over number agreement in (14).

The null form illustrated so far is not the only form whose meaning and morphosyntactic characteristics correspond to French ce/ça. This is also the case for ch in (15a) and a in (15b).

Such a variety of forms sharing the same morphosyntactic characteristics is rather surprising. One question we must ask is whether each form is a separate pronoun or whether they are allomorphs of a single morpheme. For instance, Zribi-Hertz (1994: 469) proposes that colloquial French, which she calls français avancé, possesses only one neuter pronoun, ça, and that c’ is its reduced version before vowels. The examples in (16), drawn from the Paradis corpus of Chicoutimi-Jonquière (Québec) French, illustrate this pattern.

In what follows, I will propose that the three forms introduced above, ch’, a[6], and the null form, correspond to two different pronouns, ch and a, and that the null form is an allomorph of a. Furthermore, I will identify the linguistic factors that govern the selection of each form. First, however, I consider the origin of these three forms.

4. Origin of the neuter pronouns in Picard

The connection between French ce and Picard ch’ is obvious, given the regular correspondence between French /s/ and Picard /ʃ/ shown in (17), making a common origin plausible. 

4.1 Comparison with Gallo-Romance languages that possess a neuter pronoun

The origin of a and Ø is less clear. Given the existence of neuter pronouns in many Gallo-Romance languages, one must consider the possibility that Picard a is cognate with those pronouns. For instance, Clédat (1883: 346) documents the form vou in the Forez/Roannais area, and Vignon (1901: 2) reports the existence of a neuter subject pronoun that is distinct from the masculine subject pronoun in many regions. Likewise, in western France, neuter forms that are phonetically similar to Picard a are attested. For instance, Rohlfs (1970: 183) observes oc/ac in Gascon, and Doussinet (1971) and Rézeau (1976) document o/ol in Poitevin/Saintongeais. (18) illustrates the neuter pronoun in Saintongeais.

Even though a common origin to all these neuter pronouns is plausible, a few facts argue against it. For one thing, while Poitevin and Charentais possess both subject and object clitic forms of the neuter pronoun, as we can see in (19), Picard lacks a neuter object clitic. Indeed, in Picard, a neuter object must be expressed through a strong pronoun that can be separated from the verb by a preposition and a negative element, as shown in (20). Furthermore, the distribution of Picard a is much more restricted than that of the corresponding pronouns in Saintongeais and Poitevin. While the latter can be used in contexts in which standard and colloquial French use an expletive il, as illustrated in (18a), Picard patterns like French and resorts to expletive il in such examples, as shown in (21). Similarly, while Saintongeais uses the neuter pronoun in subject-verb inversion constructions, as shown in (18c), Picard uses gendered pronouns, as can be seen in (22).

4.2 Comparison between Picard and French

The differences between the neuter clitic a in Picard and its counterparts in other Gallo-Romance languages make it unlikely that they share a common origin. In this section, I provide additional evidence against this analysis by showing that the distribution of the Picard pronominal forms closely mirrors that of ce/ça in French, thus raising the possibility that these similarities can be attributed to a common origin.

First, neuter subject pronouns are realized as clitics, while object pronouns are strong pronouns that occur in postverbal position in both varieties.[7] In both colloquial French and Picard, the subject form can only occur in preverbal position, can be separated from the verb only by other clitics, and cannot be stressed (cf. Morin 1979 and Zribi-Hertz 1994), while object pronouns can be separated from the verb and even stressed. This parallel is illustrated in (23) with Picard sentences and their French counterparts. 

Second, like French, and unlike Poitevin and Saintongeais, Picard uses the third person singular masculine subject clitic in constructions in which the subject is expletive, as we saw in (21). 

While the Picard contrast between a/Ø and ch’ that we will see below has no parallel in French, a comparison of the combined morphemes with French ce/ça reveals striking similarities. A complete comparison of these parallels and the identification of possible differences exceeds the scope of this paper. However, we will see that the complex factors that govern the choice between il(s)/elle(s) ‘he/she/they’ and ce/ça in French play a very similar role in Picard.

One crucial factor in pronoun choice in French involves subject type. While it would be inaccurate to say that clitic ce/ça can only refer to inanimate referents, it is a fact that the conditions that allow animate references are very restricted. In sentences in which the subject refers to an individual human being, for instance, ce/ça is found with DP predicates but not with adjectival or verbal predicates, as illustrated in (24).[8] In addition, the use of ce/ça is possible when a generic interpretation is involved, as the contrast in (25) shows. Finally, while DP predicates are compatible with ce/ça, this is not the case with bare noun predicates, as can be seen in (26).

Similarly, use of clitic ce/ça in reference to other non-human animate or inanimate referents also depends on the interpretation of the subject and the sentence. As can be seen in (27), a generic reading allows the use of ce/ça, but not a specific reading.[10]

A full analysis of all uses of the subject neuter clitics in Picard remains to be conducted. However, a preliminary examination of the data reveals that their distribution closely mirrors that observed in French. For instance, while neuter pronouns can refer to specific human referents, they can only do so when the predicate is a DP. This is shown in (28), where we observe the use of ch’ with the DP predicate un wépe and the use of il with adjectival and verbal predicates (est connu, connouot). Furthermore, Picard exhibits the same contrast between DP and bare noun predicates as French: while the former require neuter clitics, the latter require gender-marked pronouns, as seen in (29).

As in French, no such restriction exists concerning the category of the predicate if the subject receives a generic interpretation. As we can see in (30), when des piots is interpreted as referring to children in general, neuter subject clitics can combine with nominal (dz’étrangeus, édz horzins), verbal (surprind), and even partitive predicates (du souci, du plaisi).

Another way in which French and Picard neuter pronouns behave similarly involves the lack of agreement observed between predicates and their subjects: in both languages, verbal and adjectival predicates bear default masculine singular features. As we have seen in (14) and (15), plural subjects combine with singular verbs. Similarly, (11b) shows that with a neuter subject clitic, adjectival predicates referring to feminine subjects fail to agree with their subjects.

The coexistence of neuter pronouns and their masculine and feminine counterparts allows both languages to use one or the other in order to convey semantic differences. In (31), the author discusses how Picard cartoons that have been published are original and appreciated by their readers and uses the plural pronoun i. This sentence discusses specific cartoons. On the other hand, (32), which contains the neuter pronoun ch’, is a generic statement about the fact that cartoons in general are never taken seriously.[12]

4.3 Phonological and dialectal evidence in favor of the French connection

The syntactic and semantic factors discussed above support the hypothesis that French ce/ça and Picard ch’/a/Ø share a common origin. Phonological evidence in support of this hypothesis is provided by neighboring varieties of Picard, historical evidence for an intermediate form, and vowel epenthesis, a phenomenon that is observed within consonant sequences.

If the connection between ce and ch’ poses no particular challenge, given the correspondence between French /s/ and Picard /ʃ/ seen in (17), the phonological connection between a/Ø and French ça/ce is less straightforward. In this case, the missing link is provided by Ledieu (1909: 60–61), who documents an aspirated pronunciation for the neuter pronoun in Démuin Picard (Amiénois): “Au commencement d’une phrase, che, placé devant est, se remplace par une forte aspiration; on dit, en effet: Est vrai, est embétant, pour ch’est vrai, ch’est embétant; mais, en articulant est les paysans font entendre une forte aspiration comme si l’on devait écrire: H’est vrai ».[14] Historical evidence that such aspiration took place in the Picard spoken in the Ponthieu/Vimeu region is found in a text dating from 1834 and in which the strong form of the neuter pronoun is spelled ho.

Even though no /h/ is heard in contemporary Vimeu Picard, two factors suggest that o is an h-aspiré word. First, no liaison is possible in an expression like tout o ‘all that’, contrary to what we observe with other vowel-initial words. Second, epenthetic [e] often occurs before this pronoun. Auger (2000, 2001) shows that vowel epenthesis in d ‘of/from’ or at the end of a verb like appéll’t ‘call.3PL’ is triggered by the need to syllabify sequences of three or more consonants that exceed the syllable structure of this language. Thus, the presence of epenthetic [e] in (34a), which is similar to that observed before heut, an h-aspiré word in (34b), suggests that o is an h-aspiré word. Similarly, the presence of an epenthetic [e] before o in (35a) is unexpected given the absence of epenthesis in (35c); however, if o behaves like an h-aspiré word, we understand better why it behaves like ch Gogneu ‘the one-eyed guy’, in (35b). Finally, variation between cha and a, as seen in (36), shows that both forms exist in local varieties of Picard, thus making the hypothesis that cho/cha may have evolved into o/a and that ch’ may have produced the null neuter clitic a plausible one.[15]

On the basis of the semantic and syntactic parallels observed between French ce/ça and Picard ch/a/Ø, the differences in the distribution of Picard neuter pronouns and their equivalents in other Gallo-Romance languages, and the existence of a plausible phonological explanation that connects a/Ø to French ça/ce, I conclude that the hypothesis of a common origin for a and ça, as well as ch’/Ø and ce, is the more plausible account of the origin of Picard’s neuter pronouns.

5. How many neuter pronouns in Picard?

Even though three different forms of the neuter pronoun are observed in Vimeu Picard, we can wonder whether these truly correspond to different pronouns or whether some or all of them may be allomorphs of the same pronoun. We have already seen that Zribi-Hertz (1994: 469) proposes that français avancé possesses only one neuter pronoun, ça, and that c’ is the reduced version of this pronoun before vowel. In this section, we will see that the three neuter pronouns share many characteristics, but that they also differ from each other in important ways. Specifically, I will show that whereas a and the null form must be analyzed as allomorphs of the same clitic, ch’ is a separate pronoun that occurs in a complementary set of syntactic constructions.

5.1 Similarities

As we saw in section 3, a characteristic of all three neuter pronouns is that, unlike their masculine and feminine counterparts, they impose default masculine singular features on the verbal and adjectival predicates with which they combine. This is illustrated in (37), where singular ch’est and a soupe combine with plural subjects and where a feminine subject cooccurs with a masculine predicate adjective.

In addition to imposing default masculine singular agreement, all three pronouns share the same strong pronoun, o, as can be seen in (38):

Another important characteristic shared by all three pronouns is the fact that they encode similar semantic interpretations that differ from those expressed by masculine and feminine pronouns. For example, in (39), the presence of a neuter pronoun imposes a generic interpretation. Furthermore, while the use of neuter pronouns in reference to human referents is typically restricted to nominal predicates, as we saw in (28), all three pronouns are possible if the sentence is interpreted as a generic, as shown in (30) for ch and a.

5.2 Differences

Beyond these similarities, data collected in written texts and oral interviews reveal a complementary distribution between the three pronouns that is conditioned by both syntactic and phonological constraints. 

First, while a and Ø occur in the same syntactic constructions, ch’ is excluded from those contexts. Specifically, a and Ø occur before adjectival, verbal, and adverbial predicates, as can be seen in (40)-(42). For its part, ch’ is required with nominal, prepositional, quantified, and clausal predicates, as well as in cleft constructions, as illustrated in (43).

The choice between the allomorphs a and Ø is governed by phonological considerations. As we can see in (40)-(42), a occurs before consonants and high vowels, the most consonant-like vowels, while the null allomorph occurs before mid and low vowels due to a prohibition against similar vowels in hiatus. This deletion/merger process also affects place names that start with a low vowel when they follow the preposition à ‘at/to’: the preposition, which is normally overt before a consonant-initial place name, see (44a), is deleted before [a] and [ɑ̃] (Dawson & Smirnova 2020: 23), cf. (44b).[17]

Table 1

Neuter a/Ø vs. ch’ in Picard

Neuter a/Ø vs. ch’ in Picard

-> See the list of tables

Table 1 summarizes the distribution of the three neuter clitics. The choice between a/Ø and ch’ is conditioned by the nature of the predicate with which it occurs. The distribution of a’s allomorphs is, for its part, conditioned by the following phonological segment.

This pattern raises two questions: why does Picard have two neuter pronouns, and why is their distribution governed by the type of predicate they combine with? As far as I know, this pattern is unique among Gallo-Romance varieties. Even though the Poitevin dialect spoken on the island of Noirmoutier also allows both ol and che as neuter pronouns, variation between the two pronouns is allowed in many syntactic constructions (Léonard 1995), which differs from the Picard pattern.

If the three neuter forms are connected with French ce/ça, the fact that Picard uses ch’ and the null allomorph with vowel-initial forms of éte, and a before verbal predicates is unsurprising given that it parallels the distribution of ce and ça in many varieties of colloquial French (e.g., c’était ‘it was’ vs. ça arrive ‘it happens’). Furthermore, the tendency to use ça rather than ce before consonant-initial forms of être in many varieties of French, including that spoken in the Vimeu area (Jean-Luc Vigneux, personal communication, 7/27/09), can explain the use of a before consonant-initial forms of éte ‘be’, as seen in (40) and (42). Thus, it is possible to attribute a and its null allomorph to a historical aspiration of /ʃ/ and its subsequent deletion and to connect the contrast between a and ch/Ø to that observed between ça and ce in colloquial French.

What this hypothesis cannot explain, however, is the coexistence of ch and the null allomorph. As we have seen, the two forms are not variants of the same variable. Rather, they occur in complementary distribution, and the type of predicate determines which neuter clitic is used. That is, contrary to what is observed in colloquial French where the choice between ce and ça is, for many speakers, determined by the phonological form of être, both Picard clitics can occur before a vowel, as shown in (46). Similarly, before consonant, the a allomorph contrasts with ch, as illustrated in (47) and (48).

6. Neuter pronouns and predication

Research by Heggie (1988), Carnie (1997), DeGraff (1998), Pereltsvaig (2001), Roy (2006), and many others has revealed striking regularities in the ways in which languages distinguish different types of predicates. Whether the difference involves different copulas as in Irish (49), Éwé (50), and Jamaican Creole, an alternation between an overt and a null copula as in Haitian Creole (51), African-American English, and many English-based creoles, or a choice between gendered and neuter subject clitics as in French (52), DP predicates often behave differently than other predicates. Whether this difference is due to different copulas or different constructions remains an unresolved issue (see Roy 2006: 13–16 for a summary). Future research into this newly-identified distinction in Picard, which in addition to a difference between gendered and neuter pronouns very similar to that observed in French, also involves two different neuter pronouns, will no doubt shed valuable light on this debate.

7. Concluding remarks

Paris (1894: 166–167) and Kristol (1990: 493) both note that the opposition between masculine and neuter pronouns most likely disappeared very early in northern Gallo-Romance varieties. The fact that French and Picard both use masculine il as their expletive pronoun supports this idea. However, we have seen in this paper that Picard has not only reanalyzed the demonstrative pronoun as a neuter subject clitic like French has, but that it has also used different allomorphic forms of the neuter pronoun to create a new contrast between two neuter pronouns. The fact that this contrast does not parallel that observed between ce and ça in (colloquial) French but rather mirrors the distinction made by many languages between DP predicates and other types of predicates constitutes a morphosyntactic innovation that distinguishes the grammar of Picard from that of standard French, as well as any colloquial variety of French that I am familiar with.

The neuter subject pronouns that appear to have developed in the Vimeu variety of Picard offer a unique window into a system in which two different pronouns share the same morphological features but differ in terms of the types of predicates with which they combine. While the pattern described here characterizes the written production of authors from the Vimeu region, oral data from the same speakers and both oral and written data from speakers from the neighboring Ponthieu and Amiénois regions reveal systems in which the distinction between different predicate types plays a less categorical role than in written Vimeu Picard. The system analyzed in this paper is restricted to a small geographical area. Given the uncertain future that Picard faces as a living language and the possibility that contact with French and neighboring varieties of Picard might transform or eliminate the system described here, there is a real sense of urgency to investigate the neuter subject system in more detail and to determine what it tells us about Picard morphosyntactic structure, how these elements differ from and complement feminine and masculine subject pronouns, which semantic and pragmatic factors condition their use, and what the opposition between a/Ø and ch reveals about copular sentences in Picard in particular and in human language in general.

tours est ti pi ch’est point eutrémint
ch’est conme o et pi a n’cange point[19]

Canchon