Patois: A native geography

In this section I turn to the following statement, which I heard in Provence in 2008 from a woman then aged eighty-seven, Ms. Robert.Footnote ³ I was asking her about whether she understood what people spoke in a neighbouring village on the other side of the Rhône, a large river separating Provence from the neighbouring province of Lengadoc:

I have noted many similar items over the years, usually not paying much attention to them, as they were not what I was expecting. As researchers, we often treat those as tokens of ignorance about language matters or of alienation. Their recurrence is, however, perhaps a sign pointing to something else.

I met Ms. Robert while doing fieldwork in 2008 in Aurenja, in northern Provence. Aged eighty-seven at the time, she was the mother-in-law of one of the parents who was particularly involved in the local Occitan-language school. The utterance I reproduce above was part of a two-hour conversation we had in Occitan, a language she had not spoken since her husband had died, eleven years before. I had been asking her about whether she understood people from different localities across the Rhône. I wanted to test sociolinguist Philippe Blanchet's claim that Provençal is indeed a separate language from Occitan (or Lengadocian), his idea being that speakers themselves view those as separate (e.g. Blanchet & Schiffman Reference Blanchet and Schiffman2004). In this perspective the Rhône River would mark a possible border between both languages. Conversely, my assumption was that many differences important to linguists would not have mattered to her—undermining the claim that speakers would have viewed Provençal as a separate language. I assumed that the language on both sides of the Rhône was the same (Occitan), albeit in different dialectal forms. What I came to realize, however, was that Blanchet's argument as well as mine were probably moot. In fact, Blanchet and I probably had more in common in our approach to what language is than either of us had with Ms. Robert.

Her answers seemed to make little sense. She would acknowledge her understanding of some villages but not others, apparently randomly. Traditional speakers of patois have a documented tendency to inflate the degree of difference between their own speech and neighbouring villages, and to minimize that distance with locations further away (see Bert & Costa Reference Bert, Costa, Llamas and Watt2014). But in this case, when I mentioned locations that to me seemed nearby she would often say she did not know. She understood the Montpelhièr variety because, she said, her son-in-law's father lived there—a man she knew and understood. Even though for many places between Aurenja and Montpelhièr she could not answer, in Barcelona (in Catalonia, nearly 500 km away) she knew she had understood the patois (i.e. Catalan) when she visited.

What was gradually emerging, I realized much later, was a pattern in which she would only know what she understood if she had had direct experience with a place or with someone from that area. Whereas it is part of the imaginary of modern nation states to presume linguistic continuity based on one's representation of territory (Anderson Reference Anderson1983), Ms. Robert would make no hypothesis about places she'd never visited, not mapping language onto charted territory. Thus she appeared to contradict both Blanchet's claim (that Provençal and Lengadocian were two separate languages) as well as my own (that both varieties were part of the same language, Occitan), suggesting that something else might be at play.

Patois as an index of social domination, and more

The literature on the patois of France is abundant. One notable observation about this body of knowledge, whether descriptive or critical, is its focus on the relation with French: on the unity of French vs. the multiplicity of patois, on the high prestige of the one vs. the low prestige of the others, on the decline or the resistance of the patois. Little is thus said of the specific social dynamics of what speakers called patois, and of the world in which patois functioned.

Dialectologists have used the term patois for decades (Dauzat Reference Dauzat1938) to describe oral dialectal varieties and/or spoken on a microlocal geographical scale (Tillinger Reference Tillinger2013). Occitan sociolinguists have developed a large body of work showing how the term patois, and the sociopolitical practices it refers to, are the consequence of a split between ‘language’ and ‘non-language’ at the onset of the Modern era. According to Canadian sociolinguist Paul Laurendeau (Reference Laurendeau, Mougeon and Béniak1994), it is indeed between the seventeenth and late eighteenth centuries that the term patois shifts from describing a discredited ‘they-code’ in general to specifying linguistic varieties (‘a patois’, ‘the Provençal patois’, etc.). This shift thus occurs at the same time that modern languages emerge as political and philological objects (Bauman & Briggs Reference Bauman and Briggs2003), thus creating this dual channel for language practices and enabling them to be categorized as either language or non-language.

The Occitan sociolinguist Robèrt Lafont contributed greatly to the study of the reproduction of social and spatial inequality through language categorization. According to him, while linguistic discrimination is possibly universal, the patois/language dichotomy became a defining characteristic of the political relations between Paris and the Provinces in the Kingdom of France after the seventeenth century. The term patois gradually came to designate speech that ‘had no grammar, could not express everything, are bastardized forms of French and other languages’ (Lafont Reference Lafont1977:132). The designation of patois thus functions, like language, as a social relation—one of subordination and marginalization. While language conveys the weight of legitimate language and Parisian (good) society, patois came to index the microlocal and the bottom of the social scale in the provinces in connection with the worse aspects of natural life and disease (Costa Reference Costa2020): ‘The term patois thus sanctions this step [of historical domination], extending upon the entire social body … the linguistic-cultural ideology of national unity and the dominance of formal culture’ (Lafont Reference Lafont1977:133).

According to Lafont, the regime of patois entails three characteristics/stages: (a) the loss of the possibility to communicate in the dominated language outside the microlocal; (b) a switch to the dominant language among the elites; and (c) the loss of the dominated language among the rest of the population. Lafont contends that French linguists have focused on the geographic/microlocal aspect of patois, thus erasing the social domination aspects involved by the regime of patois, one that specifies social relations among local and national communities and separates social classes along language lines in the Modern era.

What characterizes this dualist regime of language and non-language is thus hierarchization and social domination, and, Lafont (Reference Lafont1965) claimed, alienation: while the Austro-Hungarian empire, to take another European example, comprised peoples who had been subjected to the Empire, Occitans had been alienated, he wrote—a difference that made it impossible for a national(ist) movement to emerge in Occitania. Lafont believes that alienation is all at once ethnic, cultural, and political: it is the projection onto language of a social class dialectic. But when the Occitan revival movement emerged in the 1850s, it emerged within a small, educated bourgeoisie, a class that had already been integrated into the French state. The revival movement was thus only able to imagine its struggle in terms of conflict metaphorically: the bourgeoisie of which it was an emanation could not accept this beyond the mere realm of ideas. It had already invested too much in the development of the French state and Empire. In other words, it was too late for a national moment such as had emerged in Eastern Europe. Because of the lack of the possibility of conflict, Lafont viewed the Occitan revival movement as schizophrenic on several levels. Thus, until 1914, a schoolmaster could both be a member of the Felibrige, write poetry in Provençal, and yet still severely punish his pupils for speaking patois at school (Martel Reference Martel2005).

Lafont's analysis is clearly relevant. Alienation certainly plays a crucial part in language loss, and the French republican promise of emancipation through the dominant language acted as a fundamental counter-force to language reclamation efforts. Yet, Ms. Robert suggests that there might be more at play. Lafont, and Occitan sociolinguistics at large, like the Fishmanian tradition which they pit themselves against, remain part of a structural-functional approach to language, one that leaves little room for the exploration of modes of relationality beyond structures of power.

What Ms. Robert points to is not just social domination, but also to a particular way of doing language—to a practice that cannot be reduced to an object. In other words, to an ontological difference between language and patois. My next aim is thus to show how Ms. Robert allows us to access the traces of language practices in Europe reaching beyond the imposition of the modern language regime.

Trace and aura

The task I undertake here is more modest than the reconstruction of the structural links between the Occitan people and the French state apparatus. I am instead interested in the linguistic give and take of the everyday life of patois, an example of which is given by Ms. Robert's remark above. Pieces such as these are scraps, debris of fieldwork—of which we as fieldworkers all accumulate masses over the years: ‘not the great men and celebrated events of traditional historiography but rather the “refuse” and “detritus” of history, the half-concealed, variegated traces of the daily life of “the collective”’ (Benjamin Reference Benjamin, Eiland and McLaughin2002:ix; Howard Eiland & Kevin McLaughlin, translator's foreword). I provide more examples below, but more epistemological work is needed first to treat this type of data.

Among the scraps found in Walter Benjamin's Arcades project is this piece:

This passage is quoted and discussed at length by historian Patrick Boucheron, who views it as the most precise description of the work of the historian: ‘to grasp the thing before its aura gets hold of us is to grab its trace’ (Boucheron Reference Boucheron and Benjamin2017:23). So it goes with concepts too, especially those whose scientific vs. popular knowledge status is blurred, and patois is no exception. Having become an icon of historical and social domination and a locus of struggle, it has taken possession of us and veils our understanding of the world in which patois made sense. It is thus crucial to reclaim those scraps of the social life of patois which have reached us as exactly this: traces, not just of conflict between centre and periphery, but also of a social relation, or rapport. We are left with scraps as traces, that is, clues or indices, of what social relations the world of patois prescribed and specified, and of how those differ from those in the world of language: ‘traces that may be infinitesimal make it possible to understand a deeper reality than would otherwise be attainable’ (Ginzburg Reference Ginzburg1979:280). I give further examples in the next section.

Ontology, or why revitalization fails

A large part of the work of Occitan language advocates since the 1850s has consisted in seeking to convince patois speakers that what they spoke was, in fact, a language. The rationale being first that patois, as a category, is indeed the linguistic by-product of a state of domination which is furthered by the use of that very name. And second, that patois and language are mere interchangeable labels that apply to the same world-object in a purely referential manner. In this perspective, the indexicalities associated with ‘language’ and ‘patois’ are simply ones of prestige: the language itself, existing beyond its denomination, is viewed as an immutable object, to which unfortunate indexicalities and names have been attached over time through historical domination. Worse, it may have simply lost its true name (in this case Provençal, Occitan, or Lenga d’òc) and its prestigious literary history has been forgotten. The logic, then, is that speakers should be made aware that they speak a true language, a prestigious one at that, in order for them to reclaim enough pride so they may want to speak it and save it.

Occitan language advocates have in fact consistently acted as nomenclaturists, taking the view that things have inherently correct names, however misused. They have, to put it otherwise, approached folk categorizations of language as ones of ideology, ones to be dispelled by acts of ideological clarification (Kroskrity Reference Kroskrity, Reyhner and Lockard2009). This line of reasoning descends from the Aristotelian proposition that the world is indeed the same for everyone. It applies both to the faculty of language (the act of naming things) as well as to languages, as named, semiotic objects existing in the world. For indeed, while this type of reasoning usually applies to objects in the world that get named through language, there is no reason why languages may not be subjected to the same treatment once identified as entities. Thus once what people speak in Southern France becomes identified as a language along the same lines as French or German, it follows that it must be a language, not a patois. This presupposes, however, that all humans share a concept of what a language is, irrespective of culture (Harris Reference Harris1980), reducing the question at hand to one of nomination.

Yet linguistic anthropological studies have shown repeatedly that this is not the case, and that communicative practice is itself deeply embedded in cultural structure. Michael Silverstein thus remarks that while conducting fieldwork in Australia, ‘the concept of giving an equivalent to something like English language or Creole meaning was not something that ever occurred to [his aboriginal friends]’ (Van der Aa & Blommaert Reference Van der Aa and Blommaert2014). Hauck similarly reports contrasting approaches to speech among the Guaraní and the Aché in Paraguay:

‘Speech’ can thus refer to widely differing types of practices involving very diverse types of interlocutors, but such differences are usually understood in terms of cultural perspectives on an object (speaking) that remains the same for all. Proponents of the ontological turn in anthropology such as Martin Holbraad & Morten Pedersen, however, suggest otherwise since the ‘ontological turn is not so much a matter of “seeing differently”…. It is above all a matter of seeing different things’ (Reference Holbraad and Pedersen2017:6). They add:

Ontology, then, is a way to move beyond the various representations of an object taken to have some objective reality outside the practices that constitutes it. ‘Ontology’ is a complex, polysemic term. In anthropology, most approaches use an ontological perspective to account for what status different beings receive (Keck et al. Reference Keck, Regehr and Walentowitz2015). Ontologies of language are a means to question what language/speech is to those individuals we conduct our work with. An ontological approach suggests, then, that we suspend our own judgement on language and communication, instead asking: what is it in situ? Who (or what) does it entail? How is it problematized, if at all? What worlds is it a part of, and how does it connect with other components of this world?

I suggest that patois and language are, thus, not the same thing under different names but two different things; different ways of doing linguistic-ontological work—of linking speech, beings, matter. Accordingly, patois and language should not be understood as labels representing different perspectives on the same thing, echoing comparable observations made by Magnus Course (Reference Course2018:12) among Mapuche people in southern Chile. A patois differs sharply from modern conceptions of language as neutral representations of denotational code or as iconic index of group allegiance (Bauman & Briggs Reference Bauman and Briggs2003). Further, patois and language open up to different modes of material organization: different ways of assembling language and territory, different ways of composing the realm of interlocution, and different principles defining what speech is.

Following Hauck & Heurich, I thus ask not just what assumption people hold about different languages, but also what they take language to be: ‘what [is] language in a given context, for particular people?’ (Reference Hauck and Heurich2018:2)—and with what consequences, in particular for the revival movement? The ontological umbrella helps break with Aristotelian ideas about the world being the same for everyone, and to depart from culture-as-perspective (on an immutable reality). What I hope to show in the next section is how scraps of metadiscursive practices may shed light on what language advocates broke away from in order to accomplish the work of purification that comes with the transmutation of patois into language, of index to symbol.

Linguistic-ontological work in practice: Scraps of a bygone world

I recorded the scraps below over the past twenty years, starting in 2002 when I began to learn Occitan, in its Provençal manifestation. They are from Provence and from the Lyon area—while for linguists the Occitan from Provence and the Francoprovençal from the Lyon region count as separate languages, I argue that this distinction in terms of languages makes little sense to speakers of patois. Some were recorded on paper, others just stuck in my mind over the years, precisely because they apparently made little sense. These are all from northern Provence.

Before returning to (1), by far the most common and generic occurrence (I must have heard this over a hundred times over the years), I examine the other items given above. Examples (2) and (3) are also common and refer to how individuals, in my experience at least, rarely remark on their language disappearing. Stating that the language is declining is indeed a sharp understatement, one only elicited after I had asked several times, over several conversations, if the two men quoted here still spoke in the local language with anyone. In fact, they are probably the only two remaining speakers left in the village, and they would normally converse in French, if and when they met. Examples (2) and (4) are by the same person, Fernand, and may echo one another.

Fernand was a farmer all his life in the village of Tuleta, in the southern Droma area of Provence. He had been introduced to me by Joan-Pau, who features in (3), as the last remaining real speaker of Provençal in the village. One late morning we thus went over to his house, and I was introduced as a researcher who was interested in the patois. The conversation took place partly in Occitan, partly in French; I did not record it, as this was our first (and only) meeting. Fernand had not passed on patois to his son, who was also present during the conversation. Also present was his wife, Mirelha, who, Joan-Pau had claimed, also spoke patois. Both were born in 1920, and while Fernand's parents had spoken to him exclusively in patois, Mirelha's own parents had already switched to French when addressing her. She felt uneasy speaking the language, at least in front of me, and claimed she only really understood it. Fernand was not, in fact, all that interested in talking about the language, and he would not go further than to say that yes, it was less spoken now than it used to be. Given that he stated this several times, and did not seem to mind, it could be that he had indeed not given this much thought. What did, however, interest him more was agricultural practice: what used to be cultivated, who owned fields near his own. Example (4) came when I asked him when he thought people started speaking less patois. This happened, he recounted, when tractors arrived. The horse only understood patois, but when tractors were bought in, a new era began. Presumably, the tractor spoke French, or, more likely, did not care.

The generalization of tractors in Provençal agriculture corresponds to the late 1950s, about the same time as the pre-war generations would have passed. The shift from patois to French thus corresponds to a shift in technology, and to a time which sociologist Henri Mendras (Reference Mendras1984/1992) called ‘the end of peasantry’—a consequence of the final stage of rural flight in France in the 1960s.

Like Ms. Robert, his story is not about language, or about local identity. It is about what you did, and covertly, why you stopped doing it. It situates language within a broader ontology ‘that configures land, language, ancestors, and creatures who would apprehend said language’, as Jenanne Ferguson (Reference Ferguson2019:99) writes of the Sakha in Siberia. It is not the language that went away, but the world in which this particular type of language practices made sense. It is not about identity but about the types of relations that you were able to nurture, and that the advent of mechanization rendered obsolete.

Item (5) tells a similar story. It was told to me (in French) by a woman in her seventies as my colleague Michel Bert and I were investigating the vitality of Francoprovençal as part of a sociolinguistic study ordered by the regional assembly of Rhône-Alpes (now Auvergne-Rhône-Alpes), the region created around the city of Lyon. She was part of a local association that gathered weekly to speak patois. She and her husband lived on a farm with their daughter, her husband, and their newborn child. From the outside, it seemed like the perfect setting to pass on the language, given that both the interest and material conditions were present. Asked whether she would speak patois to her granddaughter, she steadfastly rejected this idea: the child would not understand her, she said. The fact that the girl was too young to speak (or even understand) any language was not important. What mattered, more likely, was that the baby did not belong to the world in which speaking patois made sense.

I end with (1), by far the most common among patois speakers. Philippe Gardy & Robèrt Lafont (Reference Gardy and Lafont1981:87) give a similar, probably stylized example of a traditional speaker replying to sociolinguists in the course of an investigation: ‘Do I have to answer in patois? Excuse me for not speaking Occitan well’.

The sharp distinction between patois and Occitan (or, for that matter, Provençal) is indeed almost universal in the Occitan south of France among speakers of patois, pointing to the fact that the essential distinction for traditional speakers was not, as Blanchet believes, an ideological axis of differentiation (Gal & Irvine Reference Gal and Irvine2019) that was part of the ‘Occitan’ vs. ‘Provençal’ (or Gascon, or any other named variety) distinction, but an axis of ontological contrast (Hauck Reference Hauck2023) that contrasted patois to language. Whether this type of opposition of language practices is limited to the French use of the term patois or restricted to contexts long disappeared is doubtful too. Indeed, in her study on Francoprovençal in Aoste, Natalia Bichurina quotes current local legislation resulting from the linguistic post-war settlement of 1946 and authorizing the use of ‘the French and Italian languages’ in the debates of municipal councils (Art. 5.1), as well as of ‘the Francoprovençal patois’ (Art. 5.2 and 5.3 of the status of the Commune of Quart). She comments that ‘there is hardly any question that [the local language] is a language in its own right; on the contrary, “the Francoprovençal patois” is opposed to the Italian and French “languages”, and is, moreover, only accepted as a mode of oral expression’ (Bichurina Reference Bichurina2019:83; my translation). Again, patois and language are conceptualized as two very different things. The next section explores how this is so, and what difference it makes.

Decontextualizing patois, producing language

Dialectological accounts of patois were written from a positivist perspective that sought to observe and describe an objective reality usually from a dialectological perspective, and thus made little space for observations of actual practices in context. Dauzat's short (Reference Dauzat1938) study on patois, for instance, deals with linguistic description and only informs us as to actual practice unintentionally—when instructing scholars on how to collect data, for example. For instance, he writes about how careful observers were fooled by informants who had misled them by stating falsely that they originated from a particular municipality—important information to dialectologists. Yet Dauzat (Reference Dauzat1938:166) tells us that those informants said this sans penser à mal (unwittingly), again pointing to a form of relation to place that eluded dialectologists.

Other traces of the ontological status of patois are found among material produced at the beginning of the Occitan linguistic movement in the 1850s, when the revivalists needed to assert their position—that what the people spoke was indeed a language, on a par with French—and to justify it. This enterprise was not without conflict, as the account below shows. Written in 1897 by Gaston Jourdanne, a radical politician and a self-appointed historian of the Felibrige, it mentions two meetings on the question of orthography in 1852 and 1853. Spelling was in fact one of the main bones of contention at the time:

What, then, did it mean to be a ‘partisan of patois’? What did those individuals promote exactly and why? Occitan language activists have explained this dichotomy in terms of diglossic complexes. The proponents of patois were alienated, unlike those in the linguistic movement who wanted to combat diglossia. While indeed partially likely, I argue that what was at stake was also the control over indexicality. The debates between the Felibres and the partisans of patois thus provide precious indications as to what it meant to speak a patois, as opposed to a language. According to Jourdanne the issue seems to have been one of scale—the village vs. the wider space of the language that the Felibres were seeking to define. One would, however, be mistaken in thinking that the partisans of patois kept to their own village and only addressed a localized village crowd. Some poets, as Jansemin from Gascony, became popular throughout the Occitan linguistic domain. The use of patois might index locality, but this characterization fails to exhaust its full pragmatic value. Jourdanne then contrasts writing ‘as one speaks’ to the composite nature of the language of the Felibres. But what exactly does it mean to claim to write as one speaks?

To investigate this question further I propose to analyse the antagonistic stances taken by the Felibres and the Marseilles poet Victor Gelu (one of the proponents of patois) in the 1840s and 1850s in Provence—a time of profound social change that saw the arrival of the railway from Paris to Avinhon and Marseilles in the early 1850s, for example. Neither Gelu or the Felibres were of popular extraction. Yet they came from very different sociogeographic backgrounds: Gelu hailed from Marseille, a bustling Mediterranean port with a strong working class and leftist tradition. It had undergone tremendous changes due to industrialization: its population surged from 148,597 in 1836 to 233,817 in 1856 (Martin Reference Martin1867:153). In contrast, the Felibres originated from the rural outskirts of Arle and Avignon, known for staunch Catholicism and conservatism. Gelu aimed to portray the struggles and suffering of Marseilles’ people, while Felibrige literature avoided social realism, leaning towards conservatism. Yet, albeit in different ways, all wrote of a world that was vanishing before them due to industrialization, technological advancement, and political upheavals stemming from the 1789 and 1848 revolutions. Gelu captured the decline of traditional working-class culture in Marseille, while Mistral later depicted the early stages of mechanization in agriculture. The Felibres’ language imitated the French model, intending to rival it in terms of antiquity, literary usage, and prestige. Their language covered a conceptual region encompassing the southern third of France and the entire Occitan (lenga d'oc) zone. Mistral's 1876 dictionary, titled Lou tresor dóu Felibrige ou dictionnaire provençal-français, included many dialectal forms from all of the regions of Southern France, creating a composite language fit for literary purposes.

The meaning of words, carved in paper, thus became more secure and attached to proverbs and examples drawn from literary works from the Middle Ages to the nineteenth century. This is a language in the modern sense of the term, an entity that can be described, sung, loved, and—perhaps most importantly—clothed and shaped. In a 1852 poem entitled Bonjour en touti ‘Address to all’, Mistral described his aim to restore the former glory of a language he personified as a woman:

Language emerged as a product of transformation, untamed speech refined into cultured and controlled communication—one evolved to the status of subject (akin to a woman), yet with object-like attributes (akin to a doll). Mistral implied, however, that Provençal was already a language when they found her, only awaiting decent clothing (a standardized spelling system). This, as we see in the next section, is a far-fetched claim.

Resisting decontextualization

Against Mistral, Victor Gelu's entire work can be read as indexical tethering. He insisted that he wrote in patois, not in a language. Gelu (1806–1885) was a poet from Marseilles who gained fame for his Chansons provençales composed in the local urban variety of Occitan and first published in 1840 (a subsequent edition appeared in 1856). His focus was on language practices and on their social conditions of production. While he was himself not from the dregs of Marseilles, the social milieus that he described, he was often described as a keen observer and a witness (for example, in a portrait published in the Marseilles weekly Lou Cassaïre, 20 March 1864). Sociolinguist Philippe Gardy (Reference Gardy1986) quotes a later commentator who in 1916 noted that in Gelu's books ‘The language of Marseilles and of the suburbs is there, a corpse so well naturalized by the rough and pious hands of Gelu that it still seems alive’. Indeed, Gelu's work intimately entwined language use with its social context of production. In the foreword to the 1840 edition (reproduced in 1856), he stated:

Hence, a pronounced contrast emerges between metalinguistic and linguistic usage, characteristic of diglossic scenarios. Gelu consistently underscored throughout his prefaces from 1840 and 1856 that he did not compose in ‘the language’ but in patois, a distinction he revisited and expanded upon frequently. Illustrating his linguistic decisions, he wrote in 1840:

What, then, was this language that patois was not? And, perhaps more importantly as both are closely linked, what did Gelu mean when he wrote that his heroes were Marseillais? Historian Christiane Veauvy argues that Gelu's people (peuple), like his language, is not the abstract figure of the people that emerged in the eighteenth century. Instead, he wrote about those he spent his life with:

With people as with language, Gelu favoured realism and naturalism over romanticism, preferring the contextualized and the local over the decontextualized. He leaned towards the indexical as opposed to the poetic and referential, prioritizing practice over (formal) rules, even at the expense of elegance. Unlike the Felibres’ view of language as a beautiful woman, Gelu embraced ‘the raucous voices, the harsh discourse the mechanism of which escapes all the combinations of linguistics’ (Gelu Reference Gelu1856:8). The women in his poem are in fact the opposite of beautiful—rather, their beauty derives from the ugliness that is caused by the use of uncouth language. One may, of course, argue that this is in part a realist posture, yet his insistence on departing from linguistic rule may also signal an unwitting adherence to how the people he depicted conceptualized language. In the excerpt below, he contrasted rule and practice with regard to value:

Value in this passage is linked to meaning but also to the social evaluation, or price, of discourse. His naturalism tells us perhaps something about what regime of language functioned among the working class of Marseilles: speech rooted in practice. Language indexed crafts and occupations such as those of locksmiths, candlemakers, milliners, hatters, flower girls, fishermen, and so on, as well as their conditions of living (including the pollution of water and land that came with industrialization and destroyed the very conditions of life of those people). Above all, what united the working class of Marseilles was a deep-seated hatred for those they viewed as their masters.

The indexicalities present in the speech of his subjects emplaced them socially and geographically. So did the patois-as-instrument that Gelu contrasted, in the 1856 preface to his Chansons, to the language developed by the Felibres. Their Provençal, he wrote, was a composite language that indexed nothing but confusion:

What is patois?

What, then, is this patois? The linguistic world described by Gelu above bears resemblance to that of Ms. Robert, Fernand, and the others I referred to in the first part of this article, if only in the fact that a patois is above all a practice directly linked to personal experience. It cannot (or perhaps should not) be understood as ‘an arbitrary and fungible system of representation’ (Course Reference Course2018:12). What this means is that patois cannot be autonomized as a code or be reduced to pure denotation. It cannot be abstracted from the social and material conditions of those that spoke it.

Claiming that a patois is not a language thus functions as indexical tethering: attaching it securely to practice and experience. This has a number of consequences, ontologically but also in terms of language politics:

1. (i) If speaking it is a practice, speaking patois is not linked with who one is, but with what one does.

2. (ii) It cannot be abstracted or objectified, let alone personified. It can only be represented in context, in the mouths of those who speak it. Unlike language, it cannot be decontextualized, and thus cannot be standardized. Consequently, it cannot be learnt formally (it can however be acquired through various means of socialization).

3. (iii) Variation constitutes one of its fundamental features, but only insofar as patois is fundamentally chronotopical. It implies particular times, places, and relations.

4. (iv) Being chronotopical, it cannot be mapped abstractly onto territory, as languages can. The imagined community of patois is thus closely linked with personal and collective experience. This, however, does not limit it intrinsically to the ultra-local, as dialectologists argued. Ms. Robert's example of her trip to Barcelona shows that patois enables communication across space and linguistic borders and allows for far-ranging linguistic intercomprehension. This, however, cannot be assumed but is a function of experience.

5. (v) Reviving patois makes no sense. Only languages can be thought of in such terms, but as has become clear, the passage from patois to language is neither simple nor can it be achieved merely by claiming they are the same thing.

Approaching the question of patois/language from an ontological perspective thus helps clarify the scraps examined above—why the geographic world of patois speakers was so different to that of standard French speakers, for instance, and why Ms. Robert struggled to confirm her comprehension of speech between Aurenja and Montpelhièr. Also, perhaps most importantly for the issue of language revitalization, it offers insight into why patois speakers did not perceive the decline of patois as a catastrophe. What faded throughout the nineteenth and twentieth centuries, particularly post-World War I, was an entire world, wherein patois played a pivotal role but was not the sole component. In terms of political economy, the world in which patois had value was one of specific economic value and system of relations, of modes of subsistence and movement, of territory. How, then, could it make any sense to retain patois when the world it indexed and the interlocutors it presupposed was waning?