Secret Language as a “Weapon of Defense”: The Problem of Opacity in Italian Colonial Libya

Abstract

This article examines the role of referential directness and community emblematization in the documentation of Libyan sign processes construed by Italian colonial ethnographers as secretive. I examine the key texts on these practices to show that colonial ethnographers metasemiotically framed the so-called argots of Libya in terms of what was understood to be their occulting function of hiding one’s intentions and their anti-language function of opposing established society. I show that Italian colonial-ethnological preoccupations with clarity and moral unity were articulated against the discursive background of French colonial ethnology of Algeria as well as Italian racist criminology anchored in the metaphor of relative opacity.

Shortly after the Ottoman Empire surrendered the area of present-day Libya to Italy in 1911, the Royal Oriental Institute of Napoli was put under the administration of the Italian Ministry of the Colonies. This gave the ministry an official avenue to instruct interpreters and administrators in Arabic as well as Berber, “the knowledge of the populations being linked to the ethnography and to the languages [of our Libya] in order to govern them” (Ministero delle Colonie 1918, 91). This prompted a series of studies into the social practices and vernaculars of Tripolitania, Cyrenaica, and the Fezzan led especially by Francesco Beguinot of the Royal Oriental Institute of Napoli, the most important Italian Berberologist of the colonial period, and several generations of his students, such as Tommaso Sarnelli and Gino Cerbella. Guided by their French colonial antecedents as well as Italian racist criminology, these three fascist ethnographer-linguists took an interest in documenting practices they described as argotic. The study of argots (gerghi ‘secret languages’) and argotic practices throws into relief the discursively linked genealogy of secrecy and intentionality.¹ So when, in the twilight of direct Italian colonialism in Libya, Cerbella declared that “in Colonialism, the argot acquires a political aspect” (1942, 321), he was touching upon a central colonial anxiety: the opacity of the other, the concealment of one’s potentially illicit thoughts.² As Cerbella saw it, Italy’s interests in Libya ran up against the problem of argots

since individuals who want and intend to hide themselves from justice are not at issue, but it is an entire society that wants to conceal itself [occultarsi] from another society; it is a whole primitive world that wants to withdraw itself in the face of an evolved world, to not march—almost in fear of the light—in the luminous path of civilization and of progress. (Cerbella 1942, 321)

Recent anthropological work on secrecy has shifted away from the criminological tenor of earlier accounts; rather than attending to opaqueness as the stock and trade of illicit activity, attention has shifted to social practices of concealment and revelation as well as to culturally appropriate forms of controlling and circulating information (Debenport 2010, 2015, 2019). Likewise, anthropological work on intentionality has shifted away from the monological framework in which a shared code enables the exchange of individual intentions, attending instead to the culturally variable conditions of appealing to the contents of others’ minds (Duranti 1993, 2008, 2015). However, for Italian colonial researchers, practices deviating from the clarity of some or another Standard (Agha 2015, Silverstein 1996) were construed as opaque, as concealing the thoughts expressed therein.

In this essay, I demonstrate how Italian colonial understandings of Libyan vernaculars were shaped by a political-moral imperative of clarity and how such an imperative was itself grounded in colonial and criminological logics. In Italian colonial discourse, argots hide the intentions of the speaker and cultivate antagonistic sociality.³ Such antagonisms were evaluated negatively when they were anti-colonial and positively when they were anti-Arab. By examining the key colonial research on what are described as Libyan argots, I delineate the two pragmatic preoccupations that most characterize the imperative of clarity: the occulting function of hiding one’s intentions and the anti-language function of opposing established society. As I demonstrate, these preoccupations rest on the assumptions of what Michael Silverstein dubbed “monoglot standardization” (1996), namely, the valorisation of referential directness and social emblematization, that is, the use of ‘literal’ (non-‘figurative’, non-‘jargony’) language and adherence to socially and racially differentiable norms (see Agha 2015). Deviations from supposedly literal lexical simplexes and from standard form were understood unifunctionally as secretive and thus illicit, cultivating an antisociety against the colonially legitimated society.

Argot: Secret Language and Anti-language

Argots dialectically contrast with the “referential cleanliness” (Silverstein 1996, 292) and emblematic unity of Standards. From the perspective of monoglot standardization, they obstruct reference and cultivate antagonism. Gaetano Berruto, for instance, identifies occultation and anti-language as the two functions of argot (1987, 160–61). Occultation is front and center in Carla Marcato’s definition of argot as “a mode of expressing oneself that is not comprehensible to listeners” (2016, 351), that is, nonratified overhearers. On the one hand, the occulting function may be entailed by changes in meaning or form, what H. C. Conklin (1956) terms “speech disguises.” In the example of bard speech in the Moroccan High Atlas region documented by Abderrahim Youssi (1977, 141–42), an expression is first borrowed from Arabic, but with a figurative meaning, before undergoing formal transformations: so, juqa ‘crowd’ becomes ajuwaq ‘the market’ and bəddl ‘to change’ becomes sg. avəddl pl. ivəddaln ‘clothes’. According to Youssi, the group of bards using these words, embedded in utterances that are otherwise grammatically Berber, do so to conceal the meaning of what they say to each other among the residents of the villages through which they pass. On the other hand, occultation may be entailed by code choice, that is, choosing to speak in a language that is not understood by overhearers. It is in this sense that Otto Jespersen (1925, 201–2) spoke of Romani as a “concealment-language” throughout Europe and the Caucasus, either as a full natural language or as a reduced form whose sole remaining function is, supposedly, concealment (see, e.g., Soravia 1977, 90). Italian colonial assumptions about the clarity of reference, that “threats to clarity come in ethnic-, regional-, gender-, and class-based deviations” from a Standard (Silverstein 1996, 293), precluded analyses that did not center on the occulting function of argots. By contrast, while the students that Angela Reyes observed at a Korean-run supplementary school in New York described Korean as “a ‘secret language’ that teachers could not understand” (2016, 316), Reyes examines the role of speaking Korean in voicing various figures of Koreanness, each of which differently affects how interaction is structured, thereby decentering occultation in her analysis.

The supposed anti-language function was the most serious and concerning to Italian researchers devoted to colonial designs in Libya. Michael Halliday (1976) developed the notion of “anti-language” as the code of an anti-society, one that derives from but is opposed to established society. The anti-society is “a mode of resistance” against established society, and the anti-language is the vehicle for “a conscious counter-reality, not just a subcultural variant of, or angle on, a reality that is accepted by all” (1976, 582, see 576). According to Halliday, anti-language is characterized by the use of figurative language and formal transformations, shown above to serve an occulting function, but there is also a metasemantic dimension: purely anti-language expressions cannot be straightforwardly glossed in the Jakobsonian (1971, 566; 1981, 27) and Sellarsian (1950a, 1950b) manner, that is, in which to “give the meaning” of an expression is to proffer another expression that is functionally synonymous. In such a case, the metasemantic proposition “‘rosso’ (in Italian) means ·red· (in English)” presupposes that the expression-form ‘rosso’ plays the same role in Italian that the form ‘red’ does in English. Following Sellars’s convention, form-focused citations are encased by single quotes and function-focused citations by dot quotes. What Halliday suggests is that, in many cases, anti-language expressions may be explicable but not glossable since there is no corresponding role in the legitimate society played by another expression. So, describing the anti-language of male Jordanian students, Abdullah Shunnaq (1995) observes that shiqfih ‘a piece’ means “a very pretty girl” in an objectifying sense, tamir hindiyy ‘tamarind’ means “a pretty, tanned female student,” and ibn Farīd ‘Farid’s son’ denotes students who lead a very luxurious life. The assertion is that there are no expressions in Arabic that are directly equivalent—they are explicable, but not glossable.

As I show in the following sections, French colonial Berberology and Italian racist criminology set the stage for taking supposedly argotic registers as objects of study; this provided a rationale for Italian colonial linguists and ethnographers to construe certain Libyan practices—such as systematic formal transformations and periphrasis—as opaque and so both occulting and oppositional. In the terms of the colonial metadiscourse, the argots under analysis involved the production of deviant expression-forms relative to those of a Standard (see Agha 2015, 320–23)—in this case, a Libyan Arabic Standard and a number of Berber vernacular Standards. Also, Halliday’s characterization that the anti-language function of an argot entails a competing reality is a stronger formulation than what Italian researchers supposed. Though they did not use the term (which originated with Halliday), the anti-language function of argotic practice was, for Italian researchers, not about an alternative reality but about an alternative society, one that was anti-Arab for Berbers (Beguinot 1918) and Jews (Sarnelli 1924) or anti-colonial for Libyans as a whole (Cerbella 1942). It is in this sense that George Simmel described secrecy “as a sociological technique” (1950, 332), a pragmatics of sociation: the space of secrecy opened up by the occulting function “satisf[ies] within secrecy the impulse toward communion which the secret destroys in regard to the outside” (356).

Argots of French Colonial Algeria

In 1885, French Berberologists René Basset and Adolphe de Calassanti-Motylinski collected a word list of adərn awal ‘changing the speech’, which they characterized as “the argot of Mzab” in Algeria, describing it as a “secret language, mixed from Arabic and Berber, deriving above all from metaphors and word games” (Basset 1887, 437). This list was expanded upon by A. J. Mouliéras a decade later, for which he added that its “sole purpose is to render oneself unintelligible to outsiders that are listening to it” (1895, 31). According to these authors, to say aman ibəršan ‘black water’ instead of qahwa ‘coffee’ is to conceal one’s meaning, not only because the latter may be understandable to Arabophones, but also because the former is supposedly figurative rather than literal (Basset 1887, 438–39; Mouliéras 1895, 32; see also Stumme 1903, 17). The assumption is that a more semantically transparent description of the referent in Berber is nevertheless more referentially opaque to an Arab overhearer. This assumption is, of course, unjustified. These sorts of semantically transparent expressions are commonly documented throughout Berberophone North Africa. Nevertheless, these early colonial descriptions understood languages to be nomenclatures, in which objects in the world have “literal” names in a given language: that is, “real” things are “successfully and correctly referable-to by means of ‘literal’ lexical simplexes” (Silverstein 2001, 73). On this view, any deviation from the literal name is figurative and thus more opaque.⁴ This is especially clear in the example of asəmmad n lənzar ‘cold nose’ being used instead of aidi ‘dog’ (Basset 1887, 438–39; Mouliéras 1895, 32). Here, the idea is that dog literally refers to dogs, while, on the other hand, cold nose refers to dogs metonymically and therefore ‘figuratively’. This was all that was needed to consider the expression asəmmad n lənzar argotic; speakers wanting to be clear, it was thought, would simply say aidi instead.

Earlier, in their momentous tomes on Kabyle customs, Hanoteau and Letourneux likewise described the use of professional argots by peddlers and poets arising from the “need to communicate their thought to each other without risking being understood by outsiders” (1872, 307). For instance, they note that, in his professional argot, the peddler says afus d urin ‘a handful of esparto grass’ instead of argaz ‘man’ (308). Similarly, a poet might refer to a woman with the expression gwəzala ‘gazelle’ instead of thamǝṭṭuth ‘woman’ (309). The point here is that, like Basset and Mouliéras, Hanoteau and Letourneux view these variants as inherently less clear, less straightforwardly literal, than the simplex expressions with which they are referentially substitutable. Since we are not left with ethnographically rich accounts that might suggest how such expressions functioned in verbal play or verbal art or how they may have been more singularly meaningful in the lived experience of speakers, such equivalences have been reduced unifunctionally to occultation, the hiding of intentions. The discursive persistence of this logic was on display almost a century later when, in the summer of 1960 and the winter of 1961, in post-colonial Libya, Umberto Paradisi (1963) collected lexical material in el-Fuqah. Paradisi described as argotic the glosses he received from elicitation that did not seem to be literal and simplex, such as aman n tiṭ ‘water of eye’ for “tear ((as in crying)),” wa n temsi ‘that of fire’ for “stove,” and alum n tisənt ‘leaf/straw of salt’ for “tea.” There is no indication that Paradisi was told that these were argotic expressions by interviewees. Instead, it appears that this determination was his and was made on purely formal grounds.

Italian Racist Criminology

Perhaps the most important influence on the research of Beguinot, Sarnelli, and Cerbella with regard to argots was the analytic framework provided by turn-of-the-century Italian criminology. Italian criminologist Cesare Lombroso claimed that argots are “a species of hostility or conspiracy against outsiders [estranei]” (1894, xi), a point more fully elaborated in this tradition by Alfredo Niceforo, who characterized argot as “a language of war” (1897, 8). Niceforo, a member of the Italian school of racist criminology that generally attributed high levels of crime to some or another racial quality, wrote of argot that it

presents itself to us as a weapon of defense: the individual who takes up the conflict with the environment that encircles them, a conflict that may vary in form and in degree and from minimal and tenuous, and with no restriction on the number of persons, it may rise to be terrible, violent, against the entire society; the individual, I say, that takes up this conflict, has for a weapon the argot. (1897, 6)

For Niceforo, then, argot is “a weapon of defense that serves to obscure” (8), that is, it is a pragmatics of opaqueness entailing a threshold of intelligibility, since it “serves to hide thought, to assault, to cover … to put in communication only those who understand and are part of the group that speaks it” (13). In other words, it offers “the ability to communicate thought between the argotiers [i.e., argot-speakers]” (1912, 192) while remaining opaque to those beyond the threshold defined by it. In fact, Niceforo would go on to say that the argot of the people—that is, the vernacular—serves “to fight in the struggle against men of the privileged classes, to attack them, to insult them, to fool them” (78). For this reason, Walter Benjamin lauded him for recognizing “the function of argot (in the broader sense of the term) as an instrument in the class struggle” (2002, 75). Ultimately, what Niceforo elaborates is a theory of the weaponization of opaqueness

in a nuanced chain of languages moving from the clear and normal language, universal patrimony of all men and recorded in the dictionary of the language … to the pseudo-argots, or the specialization of language becoming a veritable obscurity—ending with the authentic argots where the obscurity becomes the most thick and complete, jealous, intentional, and premeditated. (1912, 97, emphasis added)

In other words, the normatively loaded division between the shared, legitimate code and “authentic argots” is anchored in the metaphor of relative opacity. This mirrors the criminological division between the upright citizen with nothing to hide (so to speak) and the delinquent: the criminal argot “is none other than the darkness that seeks to cover and hide the conflict or the crime” (Niceforo 1897, 174). So it is that, within this language ideological anchoring (see Gal 2005), expounded by Niceforo and taken up by Italian orientalists, argots are conflict registers that index antagonisms. The continuum between argots and the “normal language” is between obscurity and clarity or opaqueness and transparency, mapping the social continuum between antagonism and moral uniformity, or as Halliday would have it, between an anti-society and the established society.

In the context of Italian colonial Libya, researchers’ views on the nature of such antagonisms were responsive to developments in Italy’s colonial project. In particular, issues of language were directly tied to shifting colonial interests, including racializing projects that took language as their jumping-off point (see Kroskrity 2004, 501–3; 2020). Drawing on a tradition of racialized distinctions between Arabophones and Berberophones in North Africa (see Lorcin 1995; La Mattina 2020), Italians viewed linguistic divisions in Libya as indexical of racial antagonisms.

In 1915, the year that Italian forces withdrew to the coasts of Libya during a period they called “the rebellion” (1915–22), Francesco Beguinot convincingly made the case to the Ministry of Colonies that Berber was a language of fundamental importance. So, by 1918, Beguinot, “one of the leading Berberists in Europe,” was overseeing the preparation of “a series of manuals for the languages and for the material that they teach at the [Royal Oriental Institute of Napoli] and the history of the Berbers” (Ministero delle Colonie 1918, 304). The reason for the importance of Berber, according to Beguinot, was that Berberophones in Zwāra and the Nafūsa highlands allied with the Italians during this early period in Italy’s colonial project (which they saw as in their own tribal and economic interests; see Ahmida [2005, 29–33]), and that this language was crucial to maintaining their racial consciousness and thus what appeared to Italians as their allegiance to fellow Mediterraneans against Oriental Arabs. So, comments such as Aldo Mei’s regarding Awjila (the only Berberophone settlement in eastern Libya), that “sedentism, and their laudable care in conserving the ancestral language, have so far saved the Berber centre of Cyrenaica from Arabization and dispersion” (1914, 16), were common in the colonial period before the 1930s. From this view, Berbers were an anti-society relative to Arab dominance. Beguinot himself would argue for the importance of the Berber language to Italian colonial designs by characterizing it as a weapon of defense against Arabization, much in the manner of Niceforo:

to the process of Arabization that Tripolitania (like other regions of North Africa) has succumbed, some areas have escaped where Berber ethnic consciousness is kept alive, as is easy to see, through the many centuries of interminable conflict with the element of the Arab invader and with that Arabization … in such a state of things, use of the ancient language, itself a part and a reason for the permanence of the ethnic conscience, also represents an instrument of defense against the circumstances. (1918, 112)

However, things shifted during the 1930s. After “the rebellion” and the “re”-conquest of Tripolitania (1923–30), overlapping with the Italo-Senussi War (1923–32), the 1930s had proved to be a more secure period for fascist Italy in Libya. During this later period, one characterized by an influx of Italian settlers to Libya (Segrè 1974; Cresti 1996), the Italian colonial project was less concerned with race consciousness and more concerned with governance. Up to 1942, the central issue was not producing and reproducing racial antagonisms against Arabs, but was to consolidate colonial power, which involved incorporating Arabs into normative frameworks of colonial governance by combatting anti-colonial antagonisms (Evans-Pritchard 1946; also see Ahmida 2021). So, in the final decade of direct Italian colonialism in Libya, tribal society in Libya was taken as a dangerous anti-society in relation to colonial Italian society.

Speech Disguises

I now turn to a finer-grain description and analysis of the three most important works on argots published by Italian researchers during the colonial period in Libya: Beguinot’s 1918 account of Berber argots, Tommaso Sarnelli’s 1924 account of Tripolitanian argots, and Gino Cerbella’s 1942 account of Libyan Qurʾanic schools. First, in this section, I examine how Sarnelli and Cerbella understood the semiotics of occultation as a formal process of disguising speech (in what are sometimes referred to as “cryptolalic formations”). In the following two sections, I closely turn to Sarnelli’s and Beguinot’s explanation of the anti-language role of argots in Libya, among Libyan Jews and Libyan Berberophones, respectively. As I show, these colonial concerns, racial and criminological, reflect monoglot standardist assumptions about transparency of reference and emblematization of community.

Tommaso Sarnelli began his 1924 article “Di alcuni Gerghi Arabi della Tripolitania” (On some Arabic argots of Tripolitania) by making explicit his debt to Niceforo:

I will use ‘argot’ following the precise acceptation of Niceforo, but, also, in the more restricted sense as a category of special language “intentionally created and kept secret.” This phenomenon, as is noted, is determined by the need of a group, an association even of a small number of people—even two—to hide their own thought [il proprio pensiero] to defend themselves from the encompassing environment, more or less hostile for reasons of interests, of politics, or of simple jealousy. In such conditions, the argot is a true “weapon in the fight for life of a group; it is a protective outer-layer, and at the same time a means of offence.” (1924, 192)

The argot, in the case described by Sarnelli, involves a procedure whereby a thought, which would be clearly expressed in standard, legitimate language, is disguised by transformations to the expression-form. As Conklin (1956) would later describe it, “disguised speech” involves systematically transformed expressions which nonetheless continue to function in the same way: same meaning, different form. Sarnelli discusses two sorts of disguises, those which are written and those which are spoken. Of the first sort, used by students “to communicate among themselves without their masters’ knowledge” is gelb el-ḥurūf ‘switching the letters’ (1924, 194). In this writing disguise, through the application of transformation formulae such as ق=ل, or those in which the ʾiʿjām (i.e., the dot) is added or removed as in غ=ع or moved as in ﻴ=ﺘ (fig. 1), the expression-form ‘عليه’ becomes ‘غقتح’. So, these systematic transformational equivalences (which can be, for instance, given in a cipher) motivate possible metasemantic equations such as: “‘ظزانقش’ (in “switching the letters”) corresponds to ·طرابلس·”. The expression-forms ‘ظزانقش’ and ‘طرابلس’ both mean the same thing, but only the latter is transparent to the non-initiate, since the former is its disguised version that hides the function it plays in a given utterance. Anybody who knows how to play the Arabic language game will intuitively recognize the indexical function of ‘طرابلس’, that is, occasioning the concretization ṭrābləs ‘Tripoli’, but only those familiar with the “switching the letters” language game will recognize ‘ظزانقش’ as playing the same role. It follows that to encode an entire utterance this way is to occult the thought it expresses. The argot, here, is basically a cipher defining equivalences between spellings and concretizations as readings.

Figure 1. Letter equivalences for “switching the letters” (from Sarnelli 1924, 194). It is interesting to note that, though the expression-forms are written from right to left in conformity with Arabic orthographic conventions, the actual equational syntagmas in the examples are meant to be read from left to right (from original to disguised) in conformity with Italian orthographic conventions.

Of the disguises that apply to spoken as opposed to written language, Sarnelli also describes ziadet el-ge ‘adding the ge’ in Miṣrāta, in which a qāf (ق) is inserted between the last consonant and the final vowel so that nemshu ‘we go’ becomes nemshgew (1924, 195). This operation, like all such disguising techniques, modifies the original expression-form but not its functional role, or its meaning. So, again, this rule motivates the following metasemantic equation: “‘nemshgew’ (in “adding the ge”) corresponds to ·nemshu·”. However, it is not clear how opaque the resultant expression-form ‘nemshgew’ is in relation to its meaning, prototypically illustrated in Libyan Arabic by ‘nemshu’. This simple addition of a letter is relatively common in Arabic play-language; consider, for instance, a similar process in Yemen in which a sīn (س) is added after each syllable, so hāḏā ‘this’ becomes hāsāḏasā (Serjeant 1948, 123). In fact, Gino Cerbella claims to have heard this same operation among students in Tripoli, but with the addition of a zayn (ز) after each vocalized letter, such that ʾanā ‘I, me’ becomes ʾazanaza and ʾanta ‘you’ becomes ʾazantaza (1942, 320–21). According to Cerbella, this is enough to occult the meaning, though I believe that a richer ethnographic account may have illuminated functions other than secrecy, such as verbal play, especially among children.

In a section of his 1942 article on Libyan Qurʾanic schools, Gino Cerbella dedicates a few pages mostly to representing Sarnelli’s 1924 material while making a few connections to his own observations. As an example, Sarnelli describes an argotic practice used in al-Jifārah that involves repetition and a syllable-terminal -in, such that bayt ‘house/tent’ becomes bebin tetin (1924, 195). Cerbella mentions having heard a variation of this operation, two decades later, by a student in Tripoli (1942, 319). The variation that Cerbella heard also involved repetition but a syllable-terminal -ūn, such that jāʾa Muṣṭafā ‘Mustafa has come’ becomes jājūn ʾaʾūn mamūn ṣaṣūn ṭaṭūn fāfūn yayūn. On this basis, Cerbella hypothesized a third possible but unobserved variant in which the repeated syllable terminates in -ān (1942, 320). Among the rare novel contributions to the study of argots in his paper, Cerbella also describes an operation in which each word is terminated with an accented yāʾ (ي), for which Cerbella gives the example, Muṣṭafī jāʾī məshayī li-l-maktabī ‘Mustafa came walking to school’ (1942, 320). It is even less likely that this operation truly disguises anything much, since the original forms are quite transparent even after such a transformation and, in fact, supports my contention that many of these supposedly argotic practices had playful and creative, rather than strictly occulting, functions for the students who employed them. Nevertheless, in addition to representing Sarnelli’s material and providing his own examples of Sarnelli’s disguise types, Cerbella takes the opportunity to cite Niceforo and to emphasize that the social function of argot is as an instrument of “scoundrels and all men of ill repute” to make themselves opaque, “to elude the vigilance and the control of the personnel of the police and of the social order” (1942, 321). Consequently, he says, “this research has for us considerable value” (321).

It is worth making a few remarks here on the formal realization of the occulting function of argots. Sarnelli and Cerbella share the assumption that all deviations from a Standard are reductions in clarity, that is, referential directness in virtue of a shared code. Since, in the monological framework (critiqued by Duranti 1993, 25), communication involves appealing to a shared code as a cipher to encode and decode intentions, any aberration from this code is de jure a reduction in referential directness and hence an occultation of the speaker’s intention. Even operations, such as some of those described above, that have little effect on understanding are still understood as functioning primarily for the purposes of secrecy for this reason. And, in the case of Cerbella, writing a decade after the end of instability and of the wars in Libya, this secrecy itself was a threat to colonial interests since the anti-language function of Arabic argots was understood to be clandestine and anti-colonial in nature. However, as I show next, this anti-language function was for Sarnelli and Beguinot a matter not of anti-colonial, but of anti-Arab antagonism.

Libyan Jews and Judeo-Arabic

In the same 1924 article, Sarnelli mentions an Arabic argot used in the en-Nofliyin neighborhood of Tripoli in which the order of the consonants is shifted. Yet, to his apparent frustration, Sarnelli says that he cannot discern a clear rule as to exactly how the consonants are reordered. Unfortunately, he gives very few examples of this practice, for example, nibi nemshi li-ḥosh ‘I want to go home’ becomes binni shimni li-shoḥ (1924, 196). What is of interest here, however, is his explanation of this argot:

[It is] a language that they ironically call ʿabrāni (ancient Hebrew),⁵ perhaps deriding the Jews of ʿAmrus, who must have been its inventors for the purpose of defense and offense. The neighboring Arabs, in their turn, must have learned it to thwart their purposes. (196)

According to Sarnelli, this argotic practice was originally used as a secret language by the Jews of the neighboring Jewish quarter ʿAmrus, less than a kilometer away. He believes that it originally served an anti-language function, in Niceforo’s terms as a weapon of “defense and offense,” against their Arab neighbors. However, the counteroffensive was simply to learn the argot, neutralizing its occulting function and thus its anti-language capacity. This was because, Sarnelli believed, Arabs and Jews were natural racial antagonists.

When Nahum Slouschz had visited ʿAmrus at the turn of the century, he described it as “a network of mean, narrow alleys, bare and squalid” (1927, 35; also see De Felice 1985, 300–301) in which Jews were confined owing to “the state of relations with the Muslims” (1907, 29). However, according to Renzo De Felice, this was the product of a recent “deterioration of intercommunity relations” owing in part to the introduction of Italian banking, the relatively favorable position of Jews in European projects of modernization, and the seeds of Arab nationalism (1985, 18–19):

almost all Europeans visiting Libya at the turn of the century spoke of “racial and religious hatred” between Arabs and Jews. They may have exaggerated it and projected it back into the past, seeing it as something which had always existed. They did not realize that changes had occurred in recent decades in intercommunity relations. (24)

Two additional points are perhaps relevant. First, Libyan Jews spoke a noticeably distinct variety of Arabic referred to as Judeo-Arabic. Sumikazu Yoda suggests that separation into distinct Jewish quarters and lack of influence from literary Arabic may account, on the one hand, for linguistic innovations not present in the Libyan Arabic spoken by Muslims and, on the other, for the conservation of features that could appear archaic in relation to other Libyan Arabic dialects, such as realizing qāf (ق) as [q] rather than [ɡ] (Yoda 2018, 78–88). So, calling the practice described above—in which the consonants of words are reordered—Hebrew may have been a snarky negative comment on Jewish ways of speaking, that is, as gibberish.

Second, although not conclusive, the peddler argots recorded by researchers such as Slouschz resembled those described earlier by Basset and Hanoteau and Letourneux and not the speech disguises discussed in the previous section. That is, extant descriptions of Jewish argots in Libya foreground the figurative use of expressions, archaisms, and mixing. On this point, documenting the Arabic spoken by the Jews of Algeria, linguist Marcel Cohen described the use of argot by hagglers when confronted by police, such as bla dabar ‘silence’ from Arabic b(i)lā ‘without’ + Hebrew dabar ‘word’ (1912, 407). According to Cohen, the need for argot was possibly “more extensive when the Jews had no liberty or security during the time of Turkish domination” (405). As I similarly show in the following section with regards to Berber, where documentation exists, Judeo-Arabic argots were described as increasingly opaque with the increase of Hebrew expressions. In fact, rather than a Hebraicization of Judeo-Arabic, Slouschz preferred to think of this as simply a reduced form of Hebrew, “a few scattered and isolated words which Jews use among themselves in order that outsiders should not understand” (1927, 196; see Goldberg 1983, 90–91).

Sarnelli was clear, as was Cerbella, about the importance of his work to Italy’s colonial project in Libya: “today no colonizing people is unaware of the necessity of knowing the language, the key to the soul, of the subjects” (1924, 197). But Sarnelli’s observations not only contributed to the practical knowledge of colonial interpreters. His is another line of evidence in Italy’s “Libyan colonial archive” that Libya is plagued with antagonisms on the part of Arab aggressors which it is Italy’s duty to resolve. It is, I believe, no coincidence that Sarnelli saw things this way in the 1920s, right when it was in Italy’s interest to explain Libyan history as one of continuous and aggressive Arabization against which, at that very moment, the forces of Italian colonialism were struggling.

Berberization as De-Arabization

Francesco Beguinot’s article “II gergo dei Berberi della Tripolitania” (The argot of the Berbers of Tripolitania), based on fieldwork in the Nafūsī mountains of western Tripolitania from 1913 and 1914, ends with the observation that

In such a state of things it is understood how the use of the ancient language, if on the one hand it is a reason for the persistence of ethnic consciousness, represents also an instrument of defense against the surrounding world. (1918, 112)

Here, similar to the Hebraicizing practices of Jewish peddlers mentioned above, but unlike the speech disguises described by Sarnelli and Cerbella, Beguinot is principally interested in Berberizing practices as serving an occulting and an anti-language function. So, the article begins by remarking that what he will be calling an argot involves

a series of special words that the Berbers adopt to not be, on certain occasions, understood by Arabs, when the words are normally derived from Arabic or when the Berber words are too common and might also be known by the Arabs. (107)

In essence, what Beguinot argues is that Berberophones Berberize their Berber when in the presence of Arabs because otherwise the historically Arabized Berber they normally speak might not efficiently occult their intentions. The importance of this mechanism merits repeating: according the Beguinot, the occulting function of Berber as a language of concealment is increased as it is de-Arabized, or as Beguinot says, as it “repairs the breaches made by Arabic” (108). Consequently, as an anti-language against the wider Arab society in Libya, this use of Berber is an important means in virtue of which Berbers constitute an anti-Arab society, that is, in virtue of which they have ethnic consciousness as Berbers.

Consider, for instance, how Beguinot racializes linguistic choice alongside linguistic structure. Beguinot judges shāhi ‘tea’ to be (Libyan) Arabic: it has penetrated the Berber lexicon just as the Arabs penetrated Berber Libya, and to resist the Arab presence (and Arabization) is to repair the Arabs’ lexical breach.

Esh-shāhi is used, as in Arabic, to indicate tea. In the territory of Yafran they also say aman zaġlən=hot water. At Fassāṭo the corresponding phrase is aman iḥammāyən, which is “hot water” (from aḥammāi, qualifying adjective). The indigenes explain, for example, that some, who are wanting to go to drink tea in some place without inviting or offending the Arabs who happen to be present, would be betrayed by the use of the word esh-shāhi. (1918, 107)

The word shāhi has penetrated also [in Awjila], but [their argot] has its corresponding convention: imīn aḥmāna=hot water. Identical explanation to that which was given to me in the Jabal. (110)

Understood in metasemantic terms, Beguinot is asserting a number of functional equivalences that may likewise be given in the form of metasemantic propositions:

Each of these expression-forms corresponds to the Arabic expression-form shāhi in the sense that they all play the same role as shāhi in reference and predication, that is, they all indicate tea. In this perspective, in the order of generality, each expression-form is equally subordinated to this one referential function and for this reason they are substitutable with each other. It becomes a matter of linguistic choice. So, the Awjilan Berberophone has a choice: use the Arabic form shāhi or the Berber form imîn aḥmâna. However, a further distinction is required to understand Beguinot’s (and his Berberist predecessors’) logic. For Beguinot, the form shāhi is not only Arabic, but it is also a “literal” lexical simplex, while imîn aḥmâna is believed to be figurative (literally, ‘hot water’) and so, in some sense, contrived. (This is unjustified for a number of reasons, not least of which being that similar semantically transparent constructions are very common among Berberophones across North Africa). So, the unmarked choice would be for an Awjilan to indicate tea by shāhi, while to indicate tea by imîn aḥmâna is a motivated choice. Otherwise being functionally equal, Beguinot imagines that a Berber’s reason for deviating from the standard, unmarked shāhi is to make reference less transparent. That the alternatives to shāhi are Berber forms is, for Beguinot, the key to the maintenance of Berber race consciousness.

In another example, Beguinot does not judge the alternative expression-forms to differ in terms of what language they belong to (as in the previous example), but rather merely as a difference in referential directness owing to the literalness of reference.

To alert someone to a danger, the word ərwəl = ‘escape’ may be understood by non-Berbers; əčwa ṭārənnək = ‘raise up to your feet’ is the corresponding argot (understood at Yafran and at Fassāṭo). (1918, 108)

Again, this equivalence can be understood in metasemantic terms, except in this case the descriptor does not qualify the expression-form as belonging to one language rather than another (e.g., Berber as opposed to Arabic), but as belonging to the standard- or to the argot-dimension of the same language, in this case understood to be Berber: “‘əčwa ṭārənnək’ (in argot) corresponds to ·ərwəl·.” The point is the same, that under such an analysis the expression-forms ‘ərwəl’ and ‘əčwa ṭārənnək’ are subordinated to the same linguistic function: this time, the directive one of telling someone to run. Yet, again, the same literal-figurative distinction comes into play. The form ərwəl is taken to be literal and thus unmarked (Standard), while the form əčwa ṭārənnək is taken to be figurative (because it is analyzable into “raise (up to) foot.pl-your”) and thus less standard, less clear, less referentially direct.

The final example I discuss from Beguinot’s paper involves a continuum from most to least transparent resulting from the layered application of the literal-figurative and the Arab-Berber dichotomies. As Beguinot says: “in Fassāṭo, horse is said as sg. agmār pl. igmārən” (1918, 108). However, in argot, horse is indicated by wīn rabaʿ n iṭārən, which is analyzable as “that which has four feet.” So, in the first place, the primary division is already set up between the literal, simplex, and standard expression-form agmār and the figurative, less transparent form wīn rabaʿ n iṭārən, both subordinated to the same general function: referring to horses. Still, the reader familiar with Arabic may have noticed that, in analyzing wīn rabaʿ n iṭārən into its constituent signs, the argot form contains rabaʿ ‘four’, a paraphone of the Arabic form ʾarbaʿa ‘four’. So, for Beguinot, “a phrase in which the Arabic word indicating 4 is substituted with the phrase ‘a hand minus one’ is considered to be a grade more stretto of argot” (1918, 108). For this reason, the most argotic version of horse is not only figurative, but also purely Berberized (i.e., completely de-Arabized): fùs ġayr íǧən n iṭārən, functionally equivalent to agmār but analyzable into is constituent signs as “hand except one of feet ,” that is, “5 − 1 = 4 feet.” These considerations can be figured metasemantically as:

On this analysis, even the Berberized argot may still be Arabized, and could thus be Berberized even further. As Beguinot sees it, figurative language is not enough, a truly occulting Berber anti-language will be completely de-Arabized, that is, will be completely and purely Berber (see fig. 2). This continuum matches the one mentioned by Niceforo in which transperency of reference decends from “the clear and normal language” to the “authentic argots where the obscurity becomes the most thick and complete.” Nevertheless, a couple points are in order. Of note here is the fact that ġayr in fùs ġayr íǧen ‘hand except one’ is itself a recognizably Arabic expression-form. Yet, another important factor concerns differences in use that indexically function in ways other than occultation (which is the case, as might be clear, in most of the Italian colonial examples of argot); for instance, German orientalist Hans Stumme observed that expression-forms for numbers among the Ishelhien in southern Morocco were differentiated by gender:

for the numbers 3–29 one frequently chooses the Arabic terms [over the Berber terms]. The women and small children of the Tazĕrwalt-Shlûḥ by preference count (as far as possible) with the Berber numbers, the men by preference (from 10 up) with the Arabic ones. Therefore, the Shlûḥ call the Berber numbers laḥsâb ntimġârin, and the Arabic ones laḥsâb niirgâzĕn—i.e., women’s counting vs. men’s counting. (1899, 102, cited in Souag 2009)

While it is unclear what sociolinguistic factors may have contributed to the data Beguinot obtained, his monoglot standardist assumptions that deviations from standard form functioned as secretive, and so as counter-social resistance, foregrounded racial antagonism between Berberophones and Arabophones in Libya. His article can thus be read as an appeal to the importance of de-Arabization as a project of liberating indigenous Berbers when Italians just happened to be waging a war of conquest throughout Libya in alliance with Nafūsi Berberophones.

Figure 2. Degrees of argot; from top (literal) to bottom (figurative) and from right (Arabic) to left (Berber).

Conclusion

In this essay, I have argued that the occulting function and the anti-language function of so-called argots in Libya were products of a colonial way of seeing and hearing, one that, on the one hand, proceeded from monoglot standardist assumptions about clarity and community (Silverstein 1996; Agha 2015) and, on the other hand, from French and criminological discursive antecedents and colonial political interests (La Mattina 2020; see Burke 2014). The monoglot standardist draws the distinction between the “just right” lexical expression “of the ‘just right’ denotational Standard” (Silverstein 1996, 294) and the inevitably less transparent expressions that deviate from it. So, underlying research on argots by Beguinot, Sarnelli, and Cerbella are the presuppositions that secrecy is a matter of hiding one’s antagonistic or illicit intentions and that reductions in clarity are de jure a matter of secrecy. I have not argued that the practices described by these researchers were never used for the purposes of concealment nor that they were not told as much by their informants.

Ethnographers and linguists do encounter secrecy in a variety of ways. Aomar Boum recollects being on the other side, as party to a secret, when, in 2010 in southern Morocco, he read the graffiti on the walls of a Jewish cemetery. Out of his tour group of teachers, only he could understand the anti-Jewish graffiti written in Arabic:

As I took a couple of pictures, one of the teachers tried to ask me about the meaning of the graffiti. Before I began to translate some of the words, [the local Muslim Berber tour-guide] interrupted me in a very faint voice: “Do not tell her the exact meanings of what is written in the walls. Children play on these walls and write this nonsense. We respect local Jews.” As much as I wanted to conceal the truth from the teacher, I couldn’t, and I leaked the secret. ( 2012, 31)

What is important to recognize here is that these anti-Jewish graffiti were not intended to be secrets, they were in fact public declarations. Their meanings were only potential secrets to the other teachers because they, unlike Boum, could not make sense of them without the mediation of an Arabophone translator to “give the meaning.” The Arabicness of these graffiti was not inherently occulting. In a similar vein, Stumme once asked, “does the layperson, on average, understand a conversation between the two physicians treating him when they use the terminology of their medical profession?” (1903, 3)—that is, epistemic asymmetry presents the possibility of secrecy, but is not ipso facto occulting. In the case of (appropriately pseudonymous) San Ramón Tewa-speakers, Erin Debenport has shown how tribal members carefully manage the audiences of Tewa-languaged texts so that Tewa forms are not used inappropriately by outsiders and lose ritual efficacy as a consequence and so that rather than being pinned down, texts continue to be revisable, or “continuously perfectible” (2010, 2015). Rather than an issue of hiding intentions, such practices are ultimately about epistemic rights and propriety: if Tewa is not understandable, that is because the listener is not a ratified audience member for the Tewa-languaged discourse.

The point I am making in this conclusion is not that Beguinot’s, Sarnelli’s, and Cerbella’s research on secret languages does not stand up to scrutiny and has been superseded by more nuanced attention to cultural and pragmatic phenomenon by more recent anthropologists; it does not, and it has been. The point is, rather, to draw attention to the possible range of relationships between secrecy and languagedness (i.e., how expression-forms and utterances are understood to be tokens of linguistic-types belonging to nameable languages). The very encounter with discourse that cannot be made sense of, that is differently languaged, raises to conscious awareness the problem of opacity:

Interestingly, when confronted with a text that is written in a foreign orthography, one that we have not yet developed a familiarity with, [willed forms] of phenomenal modification are not available to us—all that is available are the “marks” themselves, which we may appreciate in terms of their various aesthetic and sensory qualities, but whose specific expressive meanings remain closed to us. In other words, while we may be able to recognize the “marks” as indications of an opaque communicative intent, that is as tokens of the communicative type “writing,” we are not able to access the specific meaning as such. (Throop 2018, 198)

As Lilith Mahmud has pointed out, when confronted with “what is hidden in plain sight, public, shared, and yet less than straightforward” (2012, 427), the anthropologist (as well as, I would contend, the linguist) aims to resolve these epistemic asymmetries, to make transparent what was opaque. In this way, social practices of “giving the meaning,” in which understanding is interactionally achieved, take on a special importance. Confronted with a text in a foreign orthography, like Arabic graffiti on a Jewish cemetery wall, we can always ask the meaning and be proffered equivalents that we do understand, though this process of coming-to-understand unfolds dialogically and this exposure might be unwelcome. The mission of Italian colonial ethnographers was to make Libya legible to and governable by Italy. Just as geographical surveys were undertaken to “domesticate the ‘unknown’ Saharan interior and transform it into a comprehensible, governable colonial domain” (Atkinson 2005, 19), linguistic surveys aimed to make sense of Libyans themselves. The most profound terrae incognitae of Libya were none other than what apparently resisted legibility and governability all together: denotationally opaque and counter-normative argots.

Under the assumption that deviations from some or another standard register, specifically those construed as such by Italian colonial linguists—such as regional varieties of Arabic (at odds with native construals of Qurʾanic or Fuṣḥā Arabic serving as the basis of such a Standard) or Berber (colonially acknowledged to be “Arabized” to various degrees)—were judged ipso facto to be deviations from clear and literal language and consequently deemed “argots” (Italian gerghi). These researchers fashioned deviations from the Standard as distinct objects of study (similarly, on “slang,” see Agha 2015), treating standard usage as the unmarked basis of direct reference and supposedly nonstandard usage as inherently motivated. Sweepingly, periphrastic and systematically transformed constructions, as well as nonstandard (or “de-Arabized”) expressions, were treated first and foremost as opaque, serving to hide one’s thoughts through occultation, to the exclusion of other possible playful, poetic, and indexical functions, or even other semiotic operations such as attenuation (see Russell 2020). It followed from criminological logic that to be anything less than clear, understood unifunctionally to obscure one’s communicative intent, was inherently antagonistic: a “weapon of defense” against Arabization (as Beguinot and Sarnelli saw it) or Italian colonialism (as Sarnelli hinted and, later, Cerbella insisted). In view of the political-moral imperative of clarity, anything deviating from a standard register is liable to be framed in terms of a semiotics of concealment.