line_gbg70.gif (2214 bytes)

Endangered Languages
line_gbg70.gif (2214 bytes)

When Last Words Become First Words: Transgressive Literacies and the Birth of Romance Textuality

by Anthony P. Espòsito
University of Pennsylvania (2006)

I. Last words Balkan style: Philology and the Bosnia Syndrome (1898).

Last words, the theme for this series of articles that comes out of last spring’s graduate conference of the same name, are somewhat disconcerting for a philologist. Philology’s traditional obsession has usually been with first words — those first and originary scribblings which initialize a culture’s, and a nation’s, textual history. Last words from a linguistic-philological perspective usually imply language death. In comparative Romance philology there is a famous instance of last words that all graduate students learn about; it is invariably told as a cautionary tale, and is meant to remind us of two things:

  1. that we always must play the hand we are dealt, that is, often we have less than perfect data; and
  2. that we must temper our conclusions in light of this less than ideal data.

The setting is the Istrian peninsula at the end of the 19th century. The two characters are the Italian linguist, Matteo Giulio Bartoli, and his informant, Antuone Udaine. Bartoli was born in 1876 in Albona d’Istria and raised within the cultural and linguistic mosaic of pre-World War I Austria-Hungary in present day Croatia. He studied historical linguistics at the University of Vienna in a rigidly neogrammarian program and in 1907 assumed the chair of linguistics at the University of Turin, a position which he held until his death in 1946. Bartoli’s early scholarly interest was the Romance language known as Dalmatian, a bridge language between the north-eastern Italian and Istro-romance dialects to its west and the Romanian dialect group in the east. At the time of Bartoli’s writing, Dalmatian was thought to be extinct, having been replaced through several waves of immigration and subsequent language contact by the more Italian-like dialects of neighboring Venezia-Friuli-Giulia in the north and west and Croatian in the south.

In 1897, Bartoli was made known of a person who claimed to be a speaker of Vegliot, a northern dialect of Dalmatian, spoken in the island of Veglia, now called Krk in Croatian. Bartoli rushed back to Istria and met Antuone Udaine Burbur, and began interviewing him, recording his vocabulary, phonology, grammar, and stories of his life. Udaine provided Bartoli with much of the information that formed the basis for his famous study published in Vienna in 1906-1907, Das Dalmatische. Bartoli’s original notes written in Italian were lost during the Second World War, though a translation of the work into Italian finally appeared in 2002. Udaine, however, was a less than ideal informant for several reasons: it came to be known that

  1. he was not really a native speaker when he revealed that he acquired the language unbeknownst to his parents, who used it as a concealment code (commonly a language used by parents when they do not want their children to understand);
  2. he was away from Krk for several extended periods and upon returning eventually became the sacristan for the local church. In this role, he acquired some knowledge of Latin, somewhat devaluing him as the ideal naive informant. Additionally, much of his language appears to suggest significant contamination from other dialects, especially Venetian and other varieties of Istro-romance. At the time of his meeting Bartoli, he hadn’t spoken Vegliot in 20 years;
  3. the Hapsburg dental plan being what it was, Udaine was toothless when Bartoli met him and his pronunciation reflects very poor dentition;
  4. advanced in age, Udaine was considerably hard of hearing. In June of 1898, Tuone Udaine, a Croatian nationalist, met a very Balkan end: he was killed when he stepped on a land mine planted by a Bosnian-Turkish separatist. Despite his complicated subject position vis-à-vis Dalmatian, Tuone Udaine’s last words uttered to and recorded by Bartoli signaled the extinction of Dalmatian romance.

However, in a way the story does not end here; last words have a way of lingering. In 1925, based in large part on his foundational work on Dalmatian, Bartoli published his famous Introduzione alla neolinguistica. Neolinguistics was, as the name suggests, a new way of looking at language, specifically linguistic change. Heretofore, historical linguistics was dominated by the neogrammarians, a Germanic school that believed in the rigid regularity of language change and the inviolate nature of the sound laws that govern it. [1] Extremely formalist and positivist in their approach, the neogrammarians viewed all mechanisms for language change as internalized and endemic to the system itself; exceptions are invariably explained away either through analogy or entropy. Having learned the lessons taught by Tuone Udaine, Bartoli saw that external factors, especially contact between languages, play as much if not a greater role in linguistic evolution. The neolinguists, also called spatial linguists, realized that social and historical circumstances can effect linguistic change as much as any internal linguistic clock. Fabiana Woodfin succinctly describes this:

Most importantly, he [Bartoli] did not believe that linguistic changes arose through internal, spontaneous evolution (also known as “parthenogenesis”), as the neogrammarians believed, but rather through contact with other idioms and languages. How does one group truly conquer another? Bartoli asked his students. By armed coercion or by making itself received with fascino? Was the prestige of a dominant group’s language truly inseparable from the prestige enjoyed by that group’s culture, institutions and world view? It is those who “give things,” Bartoli argued, who can also “give words”. (9)

The pop psychology metaphor notwithstanding, neolinguistics was a coping mechanism of sorts for radical linguistic change. The neogrammarians believed in language’s organic nature: like all organisms, languages are born and they die. For Bartoli, however, Udaine’s death did not really signal the death of Dalmatian. Musing on the supposed last speakers of Dalmatian, Prussian, and Cornish, Giovanni Bonfante writes: “On the other hand, even after the death of that ‘last speaker’, each of these languages – allegedly dead, like rabbits – goes on living in a hundred devious, hidden and subtle ways in other languages now living; the Venetian and Slavic dialects of Dalmatia, the German of the Elbe, The English of the Cornwall” (357).

This new theory of language change, that linguistic phenomena are bound to social and historical circumstances and that contacts between language groups are rarely peaceful and usually the result of conflict and struggle, had a great impact on Bartoli’s most famous student at Turin, Antonio Gramsci. It is through Bartoli’s lectures on language and his constant reaffirmation of the centrality of conflict and history, mediated by cultural seduction, or in Bartoli’s own words, fascino, that Gramsci begins to formulate his theory of hegemony and domination as the constant interplay between consent and coercion: like language itself, culture either succumbs to the allure, the fascino, of the other, or to its weapons.

It would be specious to ascribe Gramsci’s theory of language and hegemony as articulated in his Notebook 29 to Tuone Udaine’s last words. However, it does give us cause to pause. Last words are not silenced but find echo and rearticulation, and when heard or relayed, they easily become the words of others.

Philology is instinctively uncomfortable with the notion of last words, for last words signal the end of something, often a tradition. In opposite fashion, our obsession is invariably with first words. Etymological and historical dictionaries, whose task is to record the lexical history of a language, consider the first attestation of a word to be an important event. Obsessed with national origins, the search for originary and primary textual material becomes the object of the philological paper chase. It will serve us well to heed Bartoli when he writes that those in a position to give things can thus give words; the things that philology is empowered to give are the very words themselves. And if they are first words, all the better, for in this way the philologist becomes the guarantor of the nation’s origin, or at the very least, its material, textual origins.

Beware though. When we examine these initializing monuments of Romance textual culture within the frame of their material context and not just as first words that have been neatly excised and anthologized, we get a very different sense of their meaning. Far from simply being the first words of a new cultural tradition, they also attest the end of an old order. Furthermore, as acts of writing, these initializing monuments often if not always appear as appendices, glosses, and marginalia. Not texts per se but rather paratexts, these first words are also the last words written on the material document.

II. Is it not Latin? It is Devo! Linguistic Deviance in a Pre-Modern World.

I should like to offer two examples as cases in point. The first deals with Codex LXXXIX of the Biblioteca Capitolare in Verona, Italy, established in 517 A.D. as the scriptorium for the cathedral of Verona. Brought to light in 1924 by the Italian textual critic, Luigi Schiaparelli, the manuscript is of certain Mozarabic origin and contains a sequence of devotional prayers and chants associated with the Mozarabic liturgical rite, the continuation of the older Visigothic rite in Islamic occupied Spain. The codex was written in al-Andalus some time during first three decades of the 8th century in a clearly Visigothic chancery hand. From Islamic Spain, the manuscript made its way to Sardinia, then to Pisa, and at the very end of the 8th or perhaps very early 9th century, ended up in Verona. However, the importance of Schiaparelli’s discovery had nothing to do with the Mozarabic orational per se – as a genre they are common enough. What brought the manuscript to the Italian paleographer’s attention was a marginal scribble above one of the manuscript’s illustrations. Written in a different hand, late 8th – early 9th century Veronese cathedral chancery, this marginalia, known as the Indovinello veronese or Veronese Riddle, appears to have nothing to do with the text itself:

Se pareba boves alba pratalia araba & albo versorio teneba & negro semen seminaba [2]

‘he was readying the oxen[,] he was plowing the white field & was holding the white plow & sowing the black seed’

Most scholars agree on its interpretation. The riddle is likely a pen-proof, something the scribe writes to test the quill point. Performatively, the Indovinello is self-referential to the act of writing itself: the oxen are the scribe’s fingers, the white field, the parchment, the white plow, the pen and the black seed, the ink. There is more disagreement, however, surrounding the language of the Indovinello. Clearly, the morphology is quite removed from classical Latin, let alone the riddle’s imagined phonology, and because of this, most scholars comfortably affirm this text as the first written attestation of an Italian vernacular. Nonetheless, some scholars are reluctant to take the plunge and call it Italian, instead opting to classify it as an example of late 8th-century spoken or vulgar Latin. If this be the case, then it is Latin’s swan’s song. Either way, the vexing question remains: is the gloss half empty or half full?

Naming the language of the Indovinello, however, only becomes urgent when it is severed from its material containment; excised, it lacks viability unless it is grafted onto another tradition that can culturally sustain it. Regardless of what we chose to call it, the end of Latin or the beginning of Italian, what is clear when we study the Indovinello in its material context is the appearance of two parallel literacies: the established medieval Latin literacy of the Mozarabic orational, and the emergent literacy of the Indovinello, in a different hand and employing a significantly different morphology. The last words scribbled on the Mozarabic orational become the first words of something new, albeit something not easily named.

The second example brings us back to the Iberian Peninsula, the place of origin of our earlier Mozarabic codex. The texts I should like to consider are the 10th century glosses produced in the monasteries of San Millán de la Cogolla and Santo Domingo de Silos in Northern Spain. These interlinear and marginal glosses, which serve to explain difficult passages in various Latin texts, share much with the Indovinello: their interlineal or marginal inclusion within the text are the last words written on the page and like the Indovinello to Italian, they are canonized as important first words in the history of Spanish. However, unlike the Indovinello, these acts of writing are intentionally bound in meaning to the texts they seek to comment. And as acts of writing they are polyvalent, they gloss both through Latin synonymy and through translation into a quite recognizable form of primitive Spanish, or in two instances, into Basque, for whose tradition they are also first words.[3] In several cases, the glosses go beyond translatio and approach amplificatio, expanding and commenting in the vernacular on the original Latin texts. It would be worth commenting that modern editions of the glosses never do them justice as we are forced to read them in linear fashion rather than as the interlinear or marginal hypertexts that they actually are.

Why do these texts emerge when they do and take the form that they do? It is curious that I have not mentioned a French text here, purposely skirting the issue until now. When compared to Spain and Italy, writing in the vernacular in France, both in the South and in the North emerges, if you pardon the metaphor, like a baby born with a full set of teeth. The earliest examples of writing in French appear as mature texts with a clear pragmatic context. The intrusive marginality which defines both the Indovinello and the Glosas is replaced in France by a narratively sequential rendition of a treaty sworn in a French (and German) vernacular, neatly contained and introduced within the context of a Latin chronicle – the Serments de Strassbourg.[4] And there is a reason for this.

If we follow Roger Wright’s argument, the emergence of a vernacular Romance literacy is dependent on the insufficiency of the Medieval Latin writing system to represent the spoken vernaculars. According to Wright, the Medieval Latin signary, that peculiar combination of alphabetic and syllabic signs employed by early Medieval Latin scribes, came to represent two different phonological realities–one way of writing, Latin, for two ways of speaking, Latin and the vernacular. For the Spanish and Italian reader-writer at this time, the sounds that the written signs represented were far closer to his or her way of speaking than they were to anything that Cicero may have imagined. The situation in France, especially in the north, was radically different.

The need for a new way of writing, a new literacy, came with the Carolingian reforms of the 8th and 9th century. Charlemagne’s notion of the translatio imperii extended to even the linguistic realm and he saw in a recodified and reunified Latinity the single most important administrative resource available to him for the establishment of the new Rome. These spelling and pronunciation reforms, instigated by Alcuin of York, had their greatest impact in France. Because it was the center of Carolingian power, the reforms radiated from northeastern France to the periphery of the empire; and France itself was the area first foremost affected by the new linguistic reforms. This change in Latin quickly gave rise to a new linguistic consciousness. Writers throughout the Romance-speaking world realized that this new Latin, which was in essence Latin restored to its classical norms, had little to do with the language they spoke. Nowhere was this more evident than in France, where, because of early and radical diphthongization and the wide-spread loss of final unstressed syllables (not to mention a very strong Germanic presence), the spoken vernaculars were furthest removed from Alcuin’s retro-Latin.

In fact, it is likely that apocope, the loss of Latin final syllables, is the single most important reason for the need for a new way of writing in France. In Latin, most of the grammatical information of a word is contained in this final syllable. Medieval Latin scribes developed a complicated system of abbreviations that they used to represent the different suffixes of the nominal declension system and the verbal paradigm. This way of writing Latin was no doubt most incompatible for representing the vernacular in France, where, when compared to the comparatively conservative morphology of Iberian and Italian Romance, a final syllable usually looked nothing like its Latin reflex.

Vernacular literacy emerges quickly centered in France. If there are situations of competing emerging literacies in France analogous to those that I have articulated for Spain and Italy, they are between Gallo-Romance and Germanic speakers, not between French and Latin. Reflecting the Franco-Germanic bilingualism of the Carolingian center, almost all cases of vernacular glossing that incorporate French examples have a Germanic context. The Reichenau glosses produced around Lake Constance between Germany and Switzerland and the Kassel glosses written around Cologne are the most well known and studied.

III. Conclusion

I should like us to ask ourselves what is the allure, the fascino, of these last words and their relationship to this emerging vernacular literacy? Or, conversely, how does this new way of writing coerce? For me, the answer lies in its transgression. These last words on the nuclear text written by an other’s hand self-consciously celebrate their difference. As writing, they are decentered, appearing in the margins. They reject the linear structure of their Latin frame, and intrude between the lines. By virtue of their very interlinearity, they are interruptive, always reminding us that this is a new way of writing; yet, as we see in the case of the glosses, by their shared semantism, they are connective, as they join this new way of writing to the old. These last words are heterophonic, using old symbols for new sounds as they record in writing what was before just a way of saying.

Despite differences in our national languages, periods, critical dispositions, we are essentially all philologists. And as lovers of logos, we should perhaps reconsider our unease with last words. Last words have a way of both alluring and coercing us to explore new ways of writing, new ways of saying, and new ways of reading. They allure us when they remind us of the excesses and openness of language and textuality. They coerce us when by their very difference they invite us to excise them from their material context, thus effectively undoing their lastness, and instead to reorder them as those celebrated first words in a new textual imaginary.


  1. The term “blind necessity” is often used to describe the neogrammarian’s inviolate view of sound laws: “Phonetic laws, the neogrammarians dogmatically proclaim, operate with blind necessity” (Bonfante 346).
  2. I cite both the text and its tradition as found in Castellani (13). Castellani faithfully follows Schiaparelli’s own transcription except for its division into words.
  3. I offer an example of each type of gloss. All citations follow Menéndez Pidal (3-9). (1) Latin-Latin synonymy: adulterium [fornicatjonem] ‘adultery [fornication]’; (2) Latin-Romance translation: talia plura conmittunt [tales muitos fazen] ‘so many undertake it’; (3) Latin-Basque-Romance translation: precipitemur [guec ajutuezdugu] [nos nonkaigamus] ‘so that we not fall’; (4) ampliifcation and clarification: Et tertius ueniens [elo terzero diabolo uenot] ‘and the third having come [and the third devil came]’.
  4. The Serments de Strassbourg are contained in Nithard’s Latin History of the Sons of Louis the Pious (III, 5). In his chronicle, Nithard reproduces the oaths as sworn by Louis the German and Charles the Bald in French and Old High German respectively at the Treaty of Verdun (842-843). He introduces and contextualizes them by clearly announcing the shift in speaker and in language: “Lodhuvicus, quoniam maior natu erat, prior haec deinde se servaturum testatus est…” ‘Louis, being the oldest, was thus the first to swear…’ [translation mine]. Each brother then swears in the other’s language, either in ‘Romana lingua’ or in ‘Theodisca lingua’, giving French a very different type of initializing moment.


  • Bartoli, Matteo Giulio. Das Dalmatische: altromanische Sprachreste von Veglia bis Ragusa und ihre Stellung in der Apennino-Balkanischen Romania. 2 vols. Vienna: Kaiserliche Akademie der Wissenschaften, 1906; 2 vols. in 1. Nendeln (Lichtenstein): Kraus Reprint, 2005.
  • —. Introduzione alla neolinguistica. Geneva: Archivium Romanicum, 1925.
  • Bonfante, Giuliano. “The Neolinguistic Position (A Reply to Hall’s Criticism of Neolinguistics).” Language 23(1947): 344-375.
  • Castellani, Arrigo. I più antichi testi italiani. 2nd ed. Bologna: Pàtron, 1976.
  • Gramsci, Antonio. Prison Notebooks. Trans. J. A. Buttigieg & A. Callari. Ed. J. A. Buttigieg. New York: Columbia University Press, 1996.
  • Hall, Robert. “Bartoli’s ‘Neolinguistica’.” Language 22 (1946): 273-283.
  • Menéndez Pidal, Ramón. Orígenes del español. 3rd ed. Madrid: Espasa-Calpe, 1950.
  • Nithard. Histoire des fils de Louis le Pieux. Ed. Philippe Lauer. Paris: H. Champion, 1926.
  • Schiaparelli, Luigi. “Sulla data e provinenza del cod. LXXXIX Della Biblioteca di Verona (l’Orazionale Mozarabico).” Archivio Storico Italiano 7th ser. 1 (1924): 106-117.
  • Woodfin, Fabiana. “Lost in Translation: Recovering the Critical in Gramsci’s Philosophy of Praxis.” MA Thesis Boise State U, 2005.
  • Wright, Roger. Late Latin and Early Romance in Spain and Carolingian France. Liverpool: F. Cairns, 1982.
  • —. A Sociophilological Study of Late Latin. Turnhout: Brepols, 2003.



Main Menu

Created: Sunday March 16, 2008; Last updated: Sunday, April 17, 2016
Copyright © 1998, USA