1. INTRODUCTION
Verlan is a French-based code, a type of backward slang, which has been in use in the banlieues and urban lower classes of French society for the past 60 years or so. A characteristic of backward slang is the widespread use of metathesis in the coding of words, together with other processes such as truncation and resyllabification. Other languages in the world have given rise to various forms of backward slang, such as backslang in England or various urban metathetic slangs in Indonesia (see examples in Kirshenblatt-Gimblett, Reference Kirshenblatt-Gimblett1976; Vázquez Ríos, Reference Vásquez Ríos2009). What makes French verlan so distinctive is the fact that a large number of verlanised words have been lexicalised as part of normal, everyday speech, as well as the extent to which verlan is or has been used by diverse groups, mostly among young people. Many studies have described and analysed the modern forms of verlan (Bachmann and Basier, Reference Bachmann and Basier1984; Lefkowitz, Reference Lefkowitz1991; Plénat, Reference Plénat1992, Reference Plénat1995; Méla, Reference Méla1988, Reference Méla1991, Reference Méla1997 among others). Some studies analyse verlan in order to deduce more general principles applicable to the phonological structure of French (Lefkowitz, Reference Lefkowitz1989; Lefkowitz and Weinberger, Reference Lefkowitz and Weinberger1991; Azra and Cheneau, Reference Azra and Cheneau1994; Plénat, Reference Plénat1995; Walker, Reference Walker, Gess and Arteaga2006). Another group of studies considers the usage of verlanised forms in discourse and the various social settings and contexts of use of verlan (Bachmann and Basier, Reference Bachmann and Basier1984; Méla, Reference Méla1991, Reference Méla1997; Antoine, Reference Antoine1998). Modern verlan forms exclusively target content words, including monosyllables; they roughly reverse the order of syllables, often by treating monosyllables as disyllabic; and they are affected by various processes of truncation.Footnote 2
What is much less well known is the diachronic development of this form of slang, and in particular how far back it can be attested. Although the term verlan is a relatively recent coinage (c. 1950, see Lefkowitz, Reference Lefkowitz1991:52), there are anecdotal attestations of verlanised forms in the nineteenth century (see Lefkowitz, Reference Lefkowitz1991:50–52 for a summary). Only a few isolated forms are attested, possibly because it is rare to write an entire text in slang. As will be shown, most of the historical evidence for verlan prior to the Second World War in fact illustrates forms of largonji, and the few real occurrences are isolated forms which do not provide information about their context of use (see section 4.2). This article presents archival and historical evidence for the use of backward slang forms in French in the early nineteenth century. A number of codes using segment displacement or replacement, sometimes with the adjunction of various suffixes or with truncation, are attested in France in the nineteenth century. Largonji or largonji des loucherbems essentially consists in replacing the initial consonant of the word by l- and adding the initial consonant at the end of the word followed by various suffixes (see Plénat, Reference Plénat1985), as in the form largonji from jargon. Javanais is another argotic code where an infix -av- is added in between syllables, or after the first consonantic element (see Bullock, Reference Bullock1996; Plénat, Reference Plénat1991).
Verlan is generally believed to be first attested as a generalisation of a word game into a form of slang in the milieu of prisoners and convicts (Lefkowitz, Reference Lefkowitz1991:53), although the word verlan itself is only attested in 1953 and the phenomenon gains wider usage only in the second half of the twentieth century. Evidence for use of such a form of backward slang in the nineteenth century is scarce, and the first attestations of its use consist of a few isolated forms around 1840. The earliest attestations are believed to be linspré for le prince ‘the prince’ in Vidocq (Reference Vidocq1837) and Lontou for Toulon, from the name given in a letter by a convict, La Hyène, dated 1842 (Sers, Reference Sers1842:53; see also Sainéan, Reference Sainéan1912).
The document edited here is, to my knowledge, the first attested text written with features of verlan, and precedes known attestations of largonji and most early verlanised forms. It is a letter written by Benjamin Boubée, a 13-year-old student at the Catholic Seminary in Toulouse, to his friend Maurice de Guérin, later known as a writer and one of the first authors to experiment with prose poems in French literature (D'Harcourt, Reference D'Harcourt1932; Vest, Reference Vest1992). The letter is dated 10 October 1823, and comprises two parts, one written in Occitan, the regional language spoken in southern France, the other in a verlanised version of French.Footnote 3 This text using a French backward slang was not in the milieu of prisoners and convicts, but in a respectable school for gentry: the Petit Séminaire de l'Esquile was one of the best schools in Toulouse. But, while other schools, the colleges and the university tended to the liberal side, and were frequently involved in political debate and riots, the Petit Séminaire was a Catholic, conservative, and morally rigorous school (Decahors, Reference Decahors1932:52–54). Students had the status of reverends (French 'abbé'), and wore ecclesiastical attire. The local gentry who supported the royalist party and the Bourbon Monarchy all had their sons attend this school, which numbered around 500 students in 1817. Latin language and Roman antiquity, in addition to religion, formed the core of studies, and a sizeable number of students became priests (Decahors, Reference Decahors1932:54–55). At the date of the letter, Maurice de Guérin had attended the school for a year and a half.
The structure of this article is as follows. I first provide a critical edition of the letter, and a translation of it. In the second part, I provide a full glossary of verlanised forms, with their non-verlanised equivalents, all grouped according to the rules used in the transformation and the phonological shape of words. The analysis that follows examines the phonetic interpretation of the verlanised words and the question of whether this code might have been used in a spoken form, as well as its relationship with other early attestations of metathetic slangs and modern verlan. I finally conclude the article with a reappraisal of the evidence for an early formation of verlan.
2. CRITICAL EDITION OF THE TEXT
The letter is here transcribed as in the original manuscript. Deletions are indicated by strikethrough. Words that had to be restablished are indicated between square brackets. The translation provided following this edition should be understood as a free translation whose only aim is to give a sense of the content of the letter. The precise question of forms, and translation thereof, is addressed in the next section.
2.1 Critical edition of the letter
Letter by Benjamin Boubée to Maurice de Guérin, 11.X.1823, 2fol., 20x12cm. Andillac, Archives du Château du Cayla, AG471.
-
Toulouse ce 11 octobre 1823
-
La Mÿou catêto,
-
Ÿeou, catêto, t'abioi proumetut, catêto, de t'escriouré; beses bé que aou fauc. Mais ma catêto, t'abioi proumetut tabes de fa la letro en bermino latino, à counditiou, saquela Catêto, què me respoundrios dammè la mêmo bermino. Mais la mÿou catétasso, t'es bé tu mêmo apercebut qué pendent las bacansos l'on n'a pas gairé embéjétos de fa de bermino. disi pas saquèla aquos de tu, ma cateto, persoque aprep coucou es le pus balent siugè. Ainsi dounc, ma catêto tè prègi dé bouler pla récébré, (escouto pla catêto) ma letro sans bermino.
-
Ço qu'ai dit dunquos aci n'es pas qu'uno ferbento pregairo et un perdou que t'adresso le tÿou coucou dammè pla dè soumissiou et de respect.
-
Ami mon cher à moi
-
Il ÿ a d jéda longtemps que je séridais t'ericre pour te medander de tes vounelles mais moce j'ai quespre joutours tesré à la Gnanpacue je n'ai pu t'ericre tuplot. J'ai saisi le jour ou je suis nevu a Loutouse pour te métoigner mon tamiié pour toi | je n'ai pas enroce chouté la muple pour vatrailler je chate de sacher loin de moi l'ennui qui s'est enrapé de moi depuis long tonglemp. Il me darte de tenrer au Messinaire, et de p voupoir me sumer, sajer et saucer avec toi; mais je suis un peu lousagé par la sampée que tu te plais en cavances, et tursout etant joutours dans l'espoir que je vais cerevoir une letre telre de ta part ainsi donc je me tafle que tu vas tiembot me pérondre pour me lousager. je suis au moment de retarpir pour la pancagne; j'ai le a peine le temps de te prier de de me mêtre dans ta telre, que j'esrèpe cerevoir tienbot, l'adresse de Berdoulat, et dans toute sa teneur. . . Si tu lui ecris node lui mille adieux de ma [part]Footnote 4 Je te prie aussi de séprenter mes pesrects a Guival si tu le vois
-
Adieu Adieu
-
nimistre d'un néverable coucou
-
Je t'embrasse de tout mon cour
-
Benjamin Boubée
-
P.S. en cas que tu ne scussesFootnote 5 point mon adresse la voici.
-
a Mr Mr L'abbé Binjamin Boubée, place de la daurade n° 18 a Toulouse
2.2 Translation of the letter
-
Toulouse, this 11 October 1823
-
[In Occitan] My dear kitty,
-
I, kitty, had promised you, kitty, that I would write to you; you can see that this is what I do. But my kitty, I had also promised you that I would write my letter in Latin vermin, on the condition though kitty that you would answer with the same vermin. But, my pretty kitty, you must have noticed yourself that during the holidays one has little desire to do vermin? I don't say that for you, though, my kitty, because after coucou Footnote 6 you're the most diligent pupil. Thus my kitty I beg that you will kindly receive (listen carefully kitty) my letter without vermin.
-
What I have said until now is only a fervent prayer and a plea for forgiveness that your coucou sends you with all due submission and respect.
-
[In verlanised French] Dear friend of mine
-
for a long time now already I've been wanting to write to you to ask you how you are, but since I've been in the Country almost all the time I couldn't write to you any earlier. I've seized on the day I've come to Toulouse, to demonstrate to you my affection for you. I haven't yet touched my quill [in order] to work I strive to banish far from me the boredom which took hold of me a long while back. I can't wait to start back at the Seminary, and to be able to have fun, gossip, and chatter with you; but I derive some comfort from the thought that you're enjoying the holidays, and especially from my constant hope that I will receive a letter from you thus I flatter myself that you will soon reply to me to bring me comfort. I am just about to set back off into the country; I only just have time to beg you to put down for me in your letter, which I hope to receive soon, Berdoulat's address, in full. . . If you write to him, give him a thousand farewells on my behalf. I beg you also to pay my respects to Guival if you see him.
-
Farewell Farewell
-
minister of this venerable coucou
-
I embrace you with all my heart
-
Benjamin Boubée
-
P.S. in case you should happen not to know my address, here it is.
-
to Mr Mr Rev. Binjamin Boubée, no.18 place de la daurade in Toulouse
3. GLOSSARY OF THE VERLANISED FORMS
The French part of the letter comprises 288 words, 53 of which are verlanised. No monosyllabic words are verlanised, which means that out of the 89 polysyllabic words which could be verlanised, nearly 60% are. Most targetted forms are disyllabic or trisyllabic. As a rule of thumb, the first two onsets are switched, with various exceptions detailed below. A few polysyllabic words are not modified, in particular in one sentence at the end of the letter: longtemps, mêtre, l'ennui, ainsi, l'adresse, ecris, mille adieux, aussi, Berdoulat, moment, peine, teneur, prier, l'espoir. There is one instance of syntactic inversion instead of a verlanised form, in the address ami mon cher à moi, instead of the expected mon cher ami.
The word categories targeted in the verlan of the letter are similar to those targeted in largonji, javanais, or modern forms of verlan: all are content words (nouns, verbs and adverbs), possibly because these tend to be both more salient in discourse and more commonly polysyllabic than function words. Table 1 compares the frequency with which different parts of speech in the letter are targeted by onset inversion. I distinguish finite forms of verbs (V) from past participles and infinitives as these non-finite forms can also readily be used as adjectives and nouns respectively.
Table 1: Parts of speech targeted
The majority of the items targeted are verbs, which account for 54% of all words targeted. A significant proportion appear in non finite forms, as infinitives and past participles, although most of these appear in verbal constructions (in infinitival complementation or in periphrastic forms requiring the past particple of the lexical verb). This may be accounted for by the range of tenses used (predominantly past), and the use of infinitival complementation, as well as the fact that a certain number of finite verb forms, in particular auxiliaries, are monosyllables, and as such do not undergo any change in this form of verlan. No adjective occurs in a verlanised form, but the letter does not contain adjectives except in the last sentence where no word is verlanised.
This glossary presents the verlanised forms accompanied by the original French form, a translation and a part-of-speech tag, all classified by phonological shape and modification. The phonological shape of words is expressed as CV structures with indices distinguishing the order of appearance of each segment, so as to follow the structure of the modifications (C1V1.C2V2 > C2V1.C1V2, the dot noting syllable boundaries). Word final schwa is not syllabic in Standard Modern French, but it was and still is in the Southern French variety (Durand and Eychenne, Reference Durand and Eychenne2004; Eychenne, Reference Eychenne2006), where it is also regularly merged with /œ/ (Eychenne, Reference Eychenne2015). Thus words like comme /kɔ.mǝ/ are here considered as disyllabic.
3.1 Simple cases: disyllabic, trisyllabic and quadrisyllabic words without consonant clusters
Disyllabic words only present a swap of the simple consonantic onset of each syllable, as exemplified in Tables 2 and 3, and in Example 1. Consonants present in the rhyme do not play any role in the transformation (Examples 1 to 3 and Table 3).
-
(1) C1V1.C2V2C3 > C2V1.C1V2C3 joutours (twice) < toujours ‘always’ (Adv)
-
(2) C1V1.C2V2C3 > C2V1.C1V2C3 voupoir < pouvoir ‘be able to’ (V)
-
(3) C1V1C2.C3V2C4 > C3V1C2.C1V2C4 tonglemp < longtemps ‘for a long time’ (Adv)
Table 2: C1V1.C2V2 > C2V1.C1V2
Table 3: C1V1C2.C3V2 > C3V1C2.C1V2
Most trisyllabic words show inversion of the onsets of the first and second syllables, as shown for a number of words in Table 4. Example 4 shows a similar case.
-
(4) C1V1C2.C3V2.C4V3 > C3V1C2.C1V2.C4V3 pancagne < campagne ‘countryside’ (N)
Table 4: C1V1.C2V2.C3V3 > C2V1.C1V2.C3V3
The fact that only onsets are targeted by the changes means that for V-initial trisyllabic words, the onsets of the second and third syllables are swapped. This is only a minor variant on the first pattern for trisyllabic words, which still follows a generalisation that the first two onsets are switched (Table 5).
Table 5: V1C1.C2V2.C3V3 > V1C1.C3V2.C2V3
There are only two quadrisyllabic words, and they do not present any complex patterns, as only the onsets of the first two syllables are swapped: Messinaire < séminaire ‘seminary’ (N); néverable < vénérable ‘venerable’ (N).
3.2 The treatment of consonant clusters
The verlanised forms in the letter exhibit a differential treatment of onset clusters depending on the position of the syllable in the word. When the onset of the first syllable is a consonant cluster, the cluster is moved as a whole for disyllabic words (Table 6 and Example 5) and trisyllabic words alike (Table 7). There are no examples of a word containing two consecutive consonant clusters.
-
(5) C1C2V1C3.C4V2 > C4V1C3.C1C2V2 quespre < presque ‘nearly’ (Adv)
Table 6: C1C2V1.C3V2 > C3V1.C1C2V2
Table 7: C1C2V1.C3V2.C4V3 > C3V1.C1C2V2.C4V3
When the onset of the second syllable is a cluster, only the first segment takes part in the switch (Table 8). The exception concerns trisyllabic words where the first syllable does not present an onset: in that case, the first onset is moved as a whole (Example 6):
-
(6) V1.C1C2V2.C3V3 > V1.C3V2.C1C2V3 t'ericre (twice) < t'écrire ‘to write to you’ (V)
Table 8: C1V1.C2C3V2 > C2V1.C1C3V2 telre (twice) < lettre ‘letter’ (N)
3.3 Problematic cases
For some trisyllabic words, there are more complex patterns. In one case all three consonantal onsets take part in the switch, as shown in Example 7:
-
(7) C1V1.C2V2.C3V3 > C2V1.C3V2.C1V3 séridais < désirais ‘I desired, I wanted’ (V)
In a second case, the first and third onsets are swapped, potentially with a slight modification of the last vowel, as in Example 8:
-
(8) C1V1C2.C3V2.C4V3 > C4V1C2.C3V2.C1V3 Gnanpacue < campagne ‘countryside’ (N)Footnote 7
In the third example, the onset of the third syllable is not swapped with the other consonantal onset available, but moved in front of the initial vowel of the word (Example 9), to a free onset position.
-
(9) V1.C1V2.C2V3V4 > C2V1.C1V2.V3V4 tamiié < amitié ‘friendship’ (N)
In one instance only, it is the onsets of the second and third syllables that are swapped (Example 10). This instance could still be analysed as a simpler case if the targetted verb is partir ‘go’ transformed into tarpir before prefixation, as the prefix is still transparently linked to a non-prefixed form, as opposed to other prefixed forms present in the letter (cerevoir > recevoir, pérondre > répondre).
-
(10) C1V1.C2V2C3.C4V3C5 > C1V1.C4V2C3.C2V3C5 retarpir < repartir ‘return’ (Vinf)
4. UNDERLYING PRINCIPLES AND COMPARATIVE ANALYSIS
The principles used for changing the form of words in this letter are relatively simple, and amount to a process of metathesis. Contrary to what happens in modern forms of verlan, monosyllables are never verlanised.Footnote 8 Thus, there are no examples of words including significant changes to vowel quality or final vowel deletion (cf. Lefkowitz, Reference Lefkowitz1991:74, 79 for modern examples). The changes exclusively target the consonantal onsets of syllables, and switch them. In the case of complex onsets, a complex initial onset is swapped with the first consonant of the second syllable (tuplot < plutôt / plus tôt ‘earlier’), but in cases where the complex onset occurs in the second syllable, only its first consonant is moved (telre < lettre ‘letter’; tenrer < rentrer ‘come back’). Cases of complex syllable boundaries such as pesrects < respects ‘regards’ or j'esrèpe ‘I hope’ show that only onsets are moved. In some cases, the identity between the nuclei of two adjacent syllables can give the impression that the backward slang used in the letter targets whole syllables (Loutouse, joutours, cerevoir, névérable, nimistre). Only a few examples do not conform to a general rule of switching the first two onsets: all of these examples are trisyllabic. The general rule for this backward slang can be synthesized in a non-linear representation (Figure 1).
Figure 1. The general rule for the form of backward slang used in the letter
The form of verlan used in the letter can be described as a case of metathesis. Metathesis is a common phenomenon in language (see Blevins and Garret, Reference Blevins, Garrett, Hayes, Kirchner and Steriade2004), but it tends to involve adjacent segments, or liquids moving from one syllable to another (more properly called displacement; see Buckley, Reference Buckley, van Oostendorp, Ewen, Hume and Rice2011). Cases of long distance consonantal metathesis are overall quite rare, and non-systematic. In the online database of metathesis (Hume, no date), only two languages, Ayacucho Quechua and Ilocano, seem to present such metathesis, on a very occasional basis.
4.1 A written code or a phonological word game?
A word of caution is needed about the phonetic interpretation of the orthographies presented in the letter. It seems that the phonetic values to be assigned to each grapheme have to be calculated from their value in the original French word, not computed from the ordering of letters in the verlanised forms. Evidence for this comes from graphic usage for other forms of slang codes attested for French in the nineteenth century, and from the fact that attested transformations in the letter do not indicate a purely orthographic code. Forms such as j'esrèpe < j'espère clearly indicate that it is not the first consonant or group of written consonants that is switched with the second group, but the first onset of a syllable. The fact that clusters are treated differently word-initially and word-internally does not per se preclude a phonological treatment, as such differences are also attested in modern spoken forms of verlan (Plénat, Reference Plénat1995:101). Double letters are not treated as a cluster, as one would expect if the code were purely orthographic: thus lettre > telre, not *teltre. The second piece of evidence comes from the orthographic transcription used in the nineteenth century for other metathetic slangs. The wordlist of largonji studied by Plénat (Larchey, Reference Larchey1881; Plénat, Reference Plénat1985) contains a high number of words for which the parts that are untouched by transformations are left as they are normally written in the original French word, although this would normally lead to a different phonetic interpretation. Thus loucherbem is analysed as [luʃebɛm], not as [luʃɛʁbɛm] which would be its reading if the values of graphemes were not kept as in the original French form. This reflects the mid-twentieth century writing of the same word as louchébèm (Mandelbaum-Reiner, Reference Mandelbaum-Reiner1991). A comparison between the spoken largonji words reported by Mandelbaum-Reiner (Reference Mandelbaum-Reiner1991) and the phonetic interpretation of the orthographic representations analysed by Plénat (Reference Plénat1985) shows a high degree of similarity. Plénat (Reference Plénat1985) offers a number of such examples, some of which are reproduced in Examples 11 to 13:
-
(11)
-
a) lailerssem [lɛlesɛm], not [lɛlɛʁssɛm] < laisser ‘leave’
-
b) lalervem [lalevɛm], not [lalɛʁvɛm] < laver ‘wash’ (Plénat, Reference Plénat1985:80)
-
-
(12)
-
a) relagnergem [ʁǝlaɲegɛm], not *[ʁǝlaɲɛʁʒɛm] < regagner ‘win back’
-
b) relalgem [ʁǝlalgɛm], not *[ʁǝlalʒɛm] < régal ‘treat’
-
c) reliegem [ʁǝliʒɛm], not *[ʁǝljeʒɛm] < régie ‘management’
-
-
(13)
-
a) loixecres [lwazɛkʁɛs], not *[lwakskʁɛs] or *[lwaksǝkʁɛs] < croix ‘cross’
-
b) larcepes [laʁkepɛs], not *[laʁspɛs] or *[laʁsǝpɛs] < parc ‘park’
-
c) loquecem [lɔkɛkɛm], not *[lɔksɛm] or *[lɔkǝsɛm] < coq ‘cockerel’
-
Examples 11 and 12 show that the value of <er> has to be kept as [e] following its value in the original word, not the interpretation of graphemes in the modified word. Similarly, the interpretation of <g> in Example 12 follows the original value of the letter in the unmodified word. The same applies to the various orthographic representations of [k] in Example 13. The orthographic treatment of largonji in the nineteenth century leads me to posit a similar type of interpretation for the graphic representation of the form of verlan present in the letter.
The grapheme <c> normally represents [s] when followed by <e,i> and [k] when followed by <a,o,u>. Thus a form such as enroce would be read [
ʁɔsə] according to the rules of French orthography, whereas a form [
ʁɔkə] would be expected from encore [
kɔʁə] if the substitution were relevant only to sound patterns rather than orthography; in cerevoir [səʁəvwaʁ] from recevoir [ʁəsəvwaʁ] the value of <c> as [s] would be kept in any case. This interpretation is supported by the fact that in gnanpacue < campagne there seems to be an attempt at establishing a value [k] for <c>, with the problematic effect that the verlanised form would be read as [ɲ
paky] instead of the expected [ɲ
pakə]. The grapheme <s> is normally interpreted as [s] when word initial, following or preceding a consonant, and as [z] in between vowels. In examples such as séprenter, séridais, sajer, the original French word would have included [z], whereas the rules of orthography would require [s]. The opposite happens in the case of lousagé. Words such as sacher [saʃe] < chasser [ʃase] or sampée [s
pe] from pensée [p
se] illustrate the fact that the value of the consonant [s] does not change in the verlanised form, as in both positions it would be pronounced [s] according to the rules of French orthography: this is an indication that the original pronunciation must have been kept for any permuted segment. Both graphemes <s> and <c> appear in saucer < causer [koze], which according to the normal rules of French orthography would be pronounced [sose], instead of [zoke], and would be homophonous with the word saucer meaning ‘soak, wipe’, which does not seem to be an acceptable solution. The absence of some diacritics should not be taken as an indication of the precise pronunciation: diacritics were often left out in handwriting, which explains forms such as néverable < vénérable, or scusses instead of the expected sçusses or the modern susses. It seems preferable to assume that graphemes have to be interpreted with the graphemic value that they have in the original French word.Footnote
9
A separate document may offer evidence that this form of verlan was used and maybe spoken, at least by Maurice de Guérin and Benjamin Boubée. In a letter to his sister Eugénie written a year and a half before the letter in verlan, Maurice de Guérin, who had started his studies at the Petit Séminaire de l'Esquile a few days before, writes:
Mes maîtres m'aiment; mes camarades sont excellents. Je me suis lié plus particulièrement avec un dont je te parlerai. Il commence à parler ma langue, et par ce moyen nous nous communiquons l'un à l'autre, et nous jouons à la pensée sans qu'on s'en doute. (Letter, January 1822)
My teachers love me; my fellow students are excellent. I have particularly become close to one whom I will tell you about. He is starting to speak my language, and by this means we communicate with one another, and play thinking Footnote 10 without anyone noticing.
Maurice de Guérin might here be referring to the verlanised form of French attested in the letter sent to him by Benjamin Boubée. The expression ma langue strongly suggests that it is not shared by others, as French, Occitan and Latin, all known to both students and teachers at the school, would be. The first person singular possessive used cannot refer to Occitan, which at the time was either referred to as patois or by a plural possessive nòstra lenga ‘our language’. The possessive used by Maurice de Guérin indicates that he strongly identifies with this language, and that he was in possession of it before he arrived at the Petit Séminaire de l'Esquile: he has only spent a few weeks at the school, and his sister must know which language he is referring to, else he would have to specify it in his first letter to her; sans qu'on s'en doute also strongly suggests that none of the other students or staff are able to understand this code. The fact that the language is used in order to disguise a game between the two students also points to a secret language, as does the fact that it is easily learnable: his fellow student starts speaking it only a couple of weeks after Maurice de Guérin arrives at the school. It is doubtful that one of his fellow students would be capable of mastering a foreign language in such a short period of time, while a metathetic code can be learnt quite rapidly (see experiments reported in Day, Reference Day1973). Obviously, nothing indicates that the particular student he is talking about is B. Boubée, but it is a plausible leap to think he is referring to the metathetic code used in the letter. This form of backward slang might therefore be counted as a creation of Maurice de Guérin, which does not prevent it having been used by others, or recreated in other contexts.
4.2 Comparing the backward slang of the letter with early attestations of verlan
The backward slang used in this letter only targets onsets: this is reminiscent of other forms of swapping and substitutive slang attested in the second part of the nineteenth century in France, such as largonji des loucherbems (Plénat, Reference Plénat1985; Mandelbaum-Reiner, Reference Mandelbaum-Reiner1991), which also targets onsets, as does Pig Latin (Barlow, Reference Barlow2001), likewise attested in the early nineteenth century. The code used in the letter is still different from largonji in that it involves mostly metathesis, while largonji properly involves displacement and suffixation, and plays more with word boundaries (initial segment and final boundary) than syllabic structure. One should still note that in the case of vowel initial words, largonji in fact behaves much like the backward slang studied here: as is the case in the letter for forms such as j'esrèpe < j'espère ‘I hope’, the element targetted by l-replacement in largonji is not the first consonant but the first onset, as in artichaut ‘artichoke’ > [arliʃote] (Plénat, Reference Plénat1985:83; Plénat, Reference Plénat1995:102). Similarly, largonji does not include the re- prefix in the computation of the first onset of the word, as in relairefem < refaire ‘do again’ (Plénat, Reference Plénat1985:81). The form of verlan present here is more akin to the word game contrepèterie, in which a completely innocent sentence acquires or masks a secondary, secret meaning, most often with offensive or sexual content, as in Example 14:
-
(14) la cuvette est pleine de bouillon
-
“the bowl is full of broth”
-
> la buvette est pleine de couillons
-
“the bar is full of dicks”
Just like verlan and the code used in the letter, this word game targets either onsets or whole syllables, and operates a swap between two such items. But where verlan makes the swap apparent, in contrepèterie it is only underlying, and it is left to the hearer to work out what the intended (or sometimes unintentional) underlying meaning is. Contrepèterie can be traced back as far as the sixteenth century in the works of Rabelais, and the term was coined by Tabourot in 1572 (Martin, Reference Martin2005:16), but there was no other specific publication of contrepets after that before 1901. This word game is in essence a written word game, and it might have been played during the nineteenth century, as some indirect evidence suggests (Martin Reference Martin2005:18–22).
Comparison of these findings with the isolated words attested for the nineteenth century shows that there are similarities with the forms in the letter, and that most of these do not fit the present-day patterns of verlan. Thus the words attested by Vidocq (Reference Vidocq1837:254) are in fact better analysed as largonji than as verlan, as the first onset of the word is replaced by an l- and suffixed to the word: lorcefé for force ‘prison called la Force‘ and linspré for prince ‘prince’, often cited as among the first occurrences of verlan (e.g. Lefkowitz, Reference Lefkowitz1991:51) should be discarded as representing another attested type of word game, largonji. But the form of verlan attested in the letter could also account for words as common as choper ‘take, steal’ (Vidocq, Reference Vidocq1837:74). This word has traditionally been ascribed to a metaphorical use of chopper ‘stumble’: one could imagine an origin as choper < pocher ‘put into a bag, a pocket’ (TLFi), an origin which is further supported by the lexical semantics of the initial term, which seem to fit better with the new meaning of the verlanised form. On the contrary Lontou (Larchey, Reference Larchey1881:223; Sers, Reference Sers1842), or Bonbour for Bourbon and Sispi for the Pope Pie VI, attested in the 18th century, seem to fit contemporary patterns better. The form Sequinzouil for Louis XV often adduced as an early example of verlan in fact exhibits a more complex pattern that was probably only restricted to writing. Nonetheless, a form Louton, obeying the same metathetic rules as the backward slang of the letter, is also attested as designating the bagne de Toulon (La Rue, Reference La Rue and par Delaplace2011:251), sometimes alongside the variant Lontou (Bruant, Reference Bruant and de Bercy1901:425).
4.3 Linking the backward slang of the letter with modern verlan?
The backward slang attested in the letter does not obey exactly the same rules as present day verlan: instead of mainly switching syllables in polysyllabic words and swapping segments in monosyllables, the forms in the letter present only onset swaps, and as such never target monosyllables. It is still worth noting that, even if this form of verlan does not apply to whole syllables, the syllable is the main unit recognised for the substitution operations. By targeting onsets, this form of verlan highlights syllable boundaries, just as modern verlan does.
In addition, one should note that onset substitution is not unproductive in contemporary verlan usage: the noun verlan itself has been found reverlanised into lervan –mere reordering of syllables would have been inoperant in this case, because it would have reestablished the original French form l'envers ‘the reverse’ (Bachmann and Basier, Reference Bachmann and Basier1984). Furthermore, some forms of verlan, sometimes referred to as larper (<parler ‘speak’) do target the onsets of the first two syllables of a given sequence, and swap them, in contemporary usage (Sherzer, Reference Sherzer1976:24–26; Lefkowitz, Reference Lefkowitz1991:27–29). Of the various codes in usage in the nineteenth century, the form of verlan attested in the letter is thus the only one to come close to modern verlan, for it is syllable based (like javanais, but unlike word based largonji) and operates switches also marginally attested for modern verlan (as opposed to javanais which is infixal, and largonji which is based on displacement). In modern verlan, as well as in the letter, trisyllables are the problematic cases where there seems to be rule competition. Moreover, not all consonant clusters are treated in a uniform way in modern verlan (Plénat, Reference Plénat1995:101–102), and there are similarities between certain of the transformations observed for modern French monosyllabic words and for disyllabic words in the letter (see mok for comme, Plénat, Reference Plénat1995:107), with a simple consonant swap.
There is still an obvious distinction between the backward slang present in B. Boubée's letter and modern verlan: the forms in the letter seem to have been restricted at best to a very small group of students in a single school in Toulouse, while modern verlan is a widely used coding system in schools, universities and groups of young people (Azra and Cheneau, Reference Azra and Cheneau1994). As such, this restriction speaks against a direct link between the slang with features of verlan in the letter, and modern verlan.
5. CONCLUDING REMARKS
Verlan, like several other forms of backward slang, can be traced as far back as the early nineteenth century. The underlying principle of the early form of verlan attested in Benjamin Boubée's letter was not switching whole syllables, but metathesis of onsets. As such, it is a form of backward slang based on the syllabic structure of the language. The document edited here gives a much earlier and more precise dating for verlan-like codes than has been assumed until now, and provides evidence that verlanised forms were used at this time as a productive pattern for stretches of language, not only as a way of disguising a few isolated words as previous evidence appeared to indicate. Most of the early evidence for modern verlan forms appears to belong to different codes, and it seems best to speak of precursors of verlan rather than ancestors which would have been continued and expanded in the modern form of verlan.
It is significant that in the letter both Occitan and verlan are present, on a par with each other. Both languages could be understood as secret languages in the sense that they are not normally used in (formal) written form, and that Occitan was not the normal language of everyday communication between members of the gentry in the school; both languages are marginal, and when used, enhance the sense of community between students. In the early nineteenth century, the nobility departs further from Occitan, and among the gentry this language tends to be found only in extremely intimate settings, for expressing close relationships and strong emotions. As an example, Maurice de Guérin's father writes a short note in Occitan to express his disappointment at his son leaving for Paris in 1829 without having received his blessing.Footnote 11 A few years later, as a student in Paris, Maurice de Guérin finds himself part of a group of gascons who all speak Occitan, and even writes and performs a song in Occitan to thank their teacher at the graduation ceremony (Decahors, Reference Decahors1932:114; Vest, Reference Vest1992:48).
Finally, the evidence does not support an origin of verlan solely in the underworld of convicts and prisons (contra Lefkowitz, Reference Lefkowitz1991:53), but rather indicates an origin in the world of students in search of a secret language which can disguise the expression of their feelings and friendship. Even today, verlan is as much a young people's and school's code as an underworld and gang's secret language. Until more evidence can be adduced in favour of an origin in the slang forms used in prisons, this letter indicates that school jargons are as good a source for verlan and its predecessors. Furthermore, it emphasises the need for work on new archival sources to establish the origins of jargons and word games in France: as recent developments in the field of historical sociolinguistics have shown, looking at different types of evidence and sources can help further our understanding of the dynamics of language in the past (e.g. Elspaß, Reference Elspaß2005; Lodge, Reference Lodge, van der Wal and Rutten2013).
