Category Archives: Comparative Indo-European Linguistics

Parisian Conversations

A linguist look into Carolingian everyday life


The Parisian Conversations are probably the most enigmatic piece of Old High German literature that ever came down to us. What we have preserved in them is an actual and unique picture of the Carolingian day-to-day. Vernacular renderings of conversations in which people greet each other, laugh together and insult each other. The Parisian Conversation may have originated in a Early Medieval travelguide of sorts intended for Romance speaking clergymen traveling to the Franconian speaking east. It is almost as if we can see the people who uttered these sentences a thousand years ago standing on some withered Roman road, the cobble stones green with moss for the lack of maintenance, a hot wind blowing through the trees and the reed roofs of large farmhouses laying hazily in the background. A historian’s dream come true!

Strangely, however, the importance of the Parisian conversations is mostly overlooked by historians of Carolingian culture (even nowadays). Their linguistic and philological value on the other hand have been acknowledged by Old Germanicists from the nineteenth century onwards. The glosses were written down by a Romance speaking clerc, who learned Old High German as a second language and possibly heard the sentences somewhere during his travels. The “Latin” sentences that the Old High German was supposed to gloss can hardly be called Latin anymore, and is better understood as colloquial Proto-Romance. A good reminder to the historian that the clergymen writing our manuscripts did not “live” in a Latin world, they lived in a vernacular world and spoke to eachother using the Romance or the Germanic vernacular.

The sentences are concatenations of spoken language heavily influenced by the way the Romance scribe pronounced Old High German, naturally with Romance dialect features. The copyist who subsequently copied the text into the manuscript that we nowadays posses probably didn’t know any Old High German and corrupted parts of the text by making transcription errors and wrong word divisions. But still, the fact that we have an actual specimen of original non-poetic spoken Old High German makes the Parisian conversations a very valuable text for linguists.

 But not only the specialized linguist or philologist will enjoy the content of the Parisian Conversations. The vulgar and bawdy nature of quite some parts of the text make for a good laugh and a fun read. It seems very likely that some conversations were only written down for their humorous nature in the first place. In this article I want to present you a scandalous collection of sentences from the Parisian Conversations, which may together constitute a somewhat coherent converstation. I will give you the Old High German gloss, a tentative Old High German reconstruction of what the author of the text actually wanted to transcribe and the Latin translation. Finally I will give a modern rendering of the conversation as it might have looked. I will, however, not give a linguistic analysis, for that would make a whole article on itself.



The Parisian Conversations are preserved in the margins of a ninth century manuscript[1] that originated in the south of France, possibly the monastery st. Marcel at Châlons-sur-Sâone. The Parisian Conversations are copied from an older source that might be connected to the area of Sens, possibly the area where the text originated. The text is written in a dialect of Old High German which has some peculiar idiosyncracies. This has led some scholars to believe that it actually represents a dialect of Franconian spoken in the north of France, an elite Frankish sociolect surrounded by Romance speakers. Most of the idiosyncracies however can be explained as orthographical alternations and pronunciation difficulties common to the Romance speaking scribe. Please note that the Romance-Latin glosses are just as interesting to the linguist as the Old High German, for the language used there can hardly be called “Latin” anymore and truly ressembles colloquial Proto-Romance.

  Romance-Latin Old High German as glossed Reconstructed “Old High German”
83 Quot vices fottisti guanna sarden ger wanne sarden ger?
84   terue naste truwa, ne wist ech!
60 quare non fuisti ad matutinas? quandi nae guarin ger za metin wande ne waren[2] ger[3] zu metin?
61 ego nolui En valde ech[4] ne wolde!
62 tu iacuisti ad feminam in tuo lecto? ger ensclephen bitte uip in ore bette ger insliefun bit demu wif in (i)uwer bettin?
63 si sciuerit hoc senior tuus iratus erit tibi per meum caput! guez or erre az pe de semauda ger enscelphen pe dez uip so es terue u rebolgan wez (i)uwer herra daz, bi desemu (mīn)[5] hauƀda, ger insliefun so ist truwa (i)u irbolgan
64 quid dicitis vos? guaz queten ger, erra[6] waz queden ger?
65 Ausculta[7] fol[8]! coorestu, narra gahoristu, narra!
66 uelles corium de tuo equo habere in collo tuo? gualdestu abe (de)[9] tinen rose ter uht ze ine ruge? woldost du haben dīnen (h)rossen der hūt[10] zu dīnemo ruggi?
67 Stultus uolentarie fottit! narra, er sarda gerra narra, er sard gērno


I place the last two sentences which were numbered 83 and 84 in Wilhelm Braune’s edition in the same context as the conversation. I do this because sentences 83 and 84 were written on the upper margin of the manuscript page and the text seems to continue in the right margin of the page, therefore 83 and 84 seem to be followed by sentence 60 in Braune’s way of counting. The fact that 83-84 and 60-67 share a rather scandalous nature combined with the possibility that they may have followed eachother up in the original text, brings me to put the two sets of sentences together, for that may be how the text was originally intended.

                I interpret the lines as representing a verbal jousting amongst friends. The inquiry as to why one of the speakers wasn’t present at matins does not sound like something you would ask a stranger. The use of the politeness form in the second plural is used as part of their verbal game.[11] The rudeness of the retort seems to confirm this interpretation. Sentence 66 is best interpreted as referring to a physical beating by a riding whip. Sentences 83 and 84 use the word sarden which is cognate to Old Icelandic serða, which meant “unconventional intercourse”. The Latin translation fottire is the Romance continuation of Classical Latin futuere and seems to have a quite general meaning.

Reconstructed conversation

In my reconstruction of the text I will name the speakers Ruodlieb and Walthar, after the protagonists of the two most important secular poems that were written down in Carolingian times.

Ruodlieb:              How many times did you have sex?

Walthar:               Truly, I dont know!

Ruodlieb:              Why weren’t you there at matins?

Walthar:               I did not want to go!

Ruodlieb:              Did you sleep with a woman in your bed?

                               If your lord finds out that you slept, by my head, he will be so angry!

Walthar:               What did you say?

Ruodlieb:              Listen, fool!

Walthar:               Do you want to feel the skin of your horse on your back!?!

Ruodlieb:              Fool, he likes having sex (too much)!


Wolfgang Haubrichs et Max Pfister eds., „In Fracia Fui; Studien zu den romanisch-germanischen Interferenzen und zur Grundsprache der althochdeutschen ‚Pariser (Altdeutschen) Gespräche“, Abhandlungen der geistes- und sozialwissenschaftlichen Klasse 1989, 6 (Mainz 1989).

Braune, Wilhelm, Althochdeutsches Lesebuch (17th edition 1994 Tübbingen; 1875).

Wilhelm Braune, Althochdeutsche Grammatik (Tübingen 2004).

[1] the fragments of which are to be found in Cod.Vat.Reg.Lat. 566 and

[2] the indicative in /-en/ is enigmatic and looks like a dialect feature of this specific Frankish dialect

[3] it may be that ger instead of expected gi or gir is also a dialect feature of this dialect, another possibility is that is merely the Romance confusion of /i/ and /e/.

[4] See last note, ech instead of ich.

[5] I put this mīn here because it warranted by the Latin translation and to make it fall in line with similar colloquial expessions in the Middle Germanic languages and early modern English.

[6] this erra is not translated in the Latin gloss and it looks like it is quite out of place.

[7] This word is naturally the Romance word for “to listen” continued in Fr écouter, It ascoltare, Sp escuchar

[8] note here the ancestor of modern English fool!

[9] this de looks like an erroneous placement of the de in the Latin translation

[10] litterally; the skin of your horse

[11 a parallel would be the use of the politeness forms in the plays of Shakespeare

Waiting for the Barbarians

On the divide between history and historical linguistics concerning the Migration Age with the Scandinavian origin of the Goths as case study.

Sometimes neighbouring academic disciplines do not speak in the same idiom. Nowhere is this more clear than in the controversial subject of the “transformation of the Roman world”, a line of inquiry into the dynamics of the transition periode between late antiquity and the early Middle Ages favoured by leading historians such as Walter Goffart, Ian Wood and Peter Heather. It’s argument is summarized by Robert Anderson, director of the British Museum:

The period of transition between the world of late antiquity and the Middle Ages has […] traditonally been seen as one of chaos and obscurity, the “Dark Ages”. Yet modern scholarship is increasingly revealing how profoundly dynamic and influential were the cultural and intellectual shifts which mark the period. Far from initiating an age of barbarism, the successor states saw themselves as part of a Roman continuum, and readily exploited the institutions and intellectual traditions of late antiquity, adapting and reinventing them to suit their own changing circumstances and cultural traditions. (Anderson 1997; 8 )

Revisionism of this pivotal periode in Western History was not new. Since the end of the second world war scholars became mighty uncomfortable discussing the period in terms of “Germanic expansion”, the nazi discourse of “Germanic fraternity” freshly in mind. The “transformation of the Roman world” movement could be seen as an exponent of this postwar “uncomfort” and has become increasingly influential in “correcting” popular views of late antiquity and the early Middle Ages. A challenge to modern revisionism came from Bryan Ward-Perkins in 2005. Whereas the revisionists may sometimes suggest an easy and peaceful integration of Germanic peoples into a continuing and evolving Roman world, Ward-Perkins takes fault with this line of interpretation and says he is “conviced that the coming of the Germanic peoples was very unpleasant for the Roman population, and that the long-term effects of the dissolution of the empire were dramatic” (Ward-Perkins 2005: 10). His book “The fall of the Roman empire and the end of civilization” is an impressive polemic for rehabilitation of those late antique sources that speak of catastrophe, massive invasion and crumbling city walls. This is in sharp contrast to historians who dispute that potent barbarians groups marched into the Roman empire (e.g. Noble 2006; xvi) and those that argue hardly any migrations at all had taken place (Bowersock 2000:187-97).

                How does this relate to scholars working in the field of Germanic studies, who used to draw information from a wide interdisciplinary spectrum of philology, historical linguistics and folklore? For one, the divide between the disciplines has widened. Han Nijdam (2001) characterizes the revisionism of historians and folklorists after the second world war in the following way:

“suddenly there were no relics of Germanic customs or ideas anymore, neither in the Middle Ages nor in the Early Modern Period, but everything was Christian from the moment missionaries build churches and monasteries. Literary sources were often written by them and for every costum or idea a fitting bible passage could be found.” (Nijdam 2001; 60)

This assertion holds even more true for the revisionism by contemporary scholars. Whereas scholars in Germanic studies assume that along with the survival of the Old Germanic languages also other cultural products survived that could be named Germanic, scholars who work in the “transformation of the Roman world” paradigm reject such an assertion[1]. For them Germanic identity arose anew in the margins of the limes (i.e. the Roman border) and was barely inherited from prehistoric times. For them Germanic identity arose from the cultural dialectic between Roman rural society and small groups of barbarian immigrants.

For every comparative Indo-European linguist the inadequacies of such a paradigm are evident. The parallels in literary motives, metre and even specific formulas between Celtic, Greek, Vedic, Slavic and Germanic cosmogenic writings are astonishing. They are hardly explicable by any other means than assuming heritage from a common prehistoric literary reservoir, a reservoir justly called Indo-European (e.g. see Puhvel Comparative Mythology 1989). The comparative linguist may draw the same conclusions for the non-christian / non-Roman parallels in Old Germanic literary monuments e.g. the vestiges of a Common-Germanic pantheon (such as in the case of the worship of Wodan and Thunar). Common heritage from a cultural tradition which, in analogy to our linguistic terminology, may be called Proto- or Common-Germanic, seems likely. Rejections of such a theoretical construct by Noble and Goffart (2006: 12) are unjustified for they ignore empirical data mined from the field of comparative mythology and comparative law arguing in favour of it.

But to indicate the divide between historians and comparative linguists in the area of migration history I’d like to discuss the case of the Gothic migration, which is basically the case of the Scandinavian origin myth. This myth is to be found in Jordanes[2]Getica, an abridgment of an earlier work, historia gothorum, by the Gothic historian Cassiodorus recounting the history of the Goths. Jordanes wrote it at the Byzantine court when the Ostrogothic kingdom of Italy had fallen under renewed Byzantine attempts of reconquest. According to this myth, Scandinavia was the womb of peoples from which the Goths set out in three ships. From the Baltic they made their way to the Black Sea when they arrive in the scope of Roman historians. The veracity of this myth stands at the center of the problem.

The main historical polemic on this topic is between Herwig Wolfram and Walter Goffart. Herwig Wolfram is a student of Reinhard Wenskus on whose work on ethnogenesis he largely draws. Wolfram formulated a thesis later known as the Wenskus-Wolfram thesis that focusses on the leading role in ethnogenesis for so called “nuclei of tradtion” (Traditionskerne). These are ancient families whose connections to the past gave a common focus for the myriad ethnic groups within a multi-ethnic confederation. The multiethnic groups would associate and identify themselves with these ancient families and accept their tribal name (Wolfram 2006: 52-54). His argument is that although Gothic identity has been reinterpreted and adapted multiple times throughout Gothic history, Jordanes is basically right in recounting the Scandinavian origin myth. This is how the tribal name of the 2nd century Gutones in Tacitus’ Germania survived in the late antique Γότθοι of Procopius. Goffart rejects Wolfram’s arguments as reading truth in mere fiction and argues for a strict 6th century Byzantine creation ex nihilo. That this cannot be true is argued by Svennung (1967: 235) who deemed the ethnonyms cited by Jordanes to be authentic correspondances to Scandinavian tribal names.

In the field of comparative Indo-European linguistics a Scandinavian origin for the late antique Goths is widely accepted. Augustin Speyer (2007) states:

“[Das Ostgermanische ist] Ursprünglich im südlichen und südöstlichen Skandinavien beheimatet; die Inselnamen Gotland and Bornholm (< Borgundarholmr) weisen auf Bezeihung dieser Plätze zu ostgermanischen Stämmen hin.”

Jasanoff (2008) is more carefull in placing the original home of the Goths in Scandinavia, but the connection the Baltic Gutones is not in doubt. He states this view as follows:

“Like other East Germanic tribes such as the Vandals, Burgundians, Gepids and Heruls, the Goths originally lived in the area of present-day Poland and eastern Germany; their own traditions place their earliest homes in southern Sweden.”

An interesting but very late source contiguous to this Scandinavian origin myth is known to most Old Germanicists but remains undiscussed by most historians (Wenskus on the other hand treats it in some detail). This is the Gutasaga, composed around 1220 but preserved in a fourteenth century manuscript, which recounts that a third of the people left Gotland in ships and traveled to the Baltic coasts, from whence they traveled through Eastern Europe to arive in the land of the Greeks.

Sīþan af þissum þrim aucaþis fulc ī Gutlandi sō mikit um langam tīma, at land elpfti þaim ai alla fyþa. Þā lutaþu þair bort af landi huert þriþia þiauþ, sō alt sculdu þair aiga oc miþ sīr bort hafa, sum þair ufan iorþar āttu.

From these three the people on Gotland multiplied throughout such a long time that the land could not feed them all. Then they drew lots so that a third of the people left the land, so that they would have and take with them everything which they possessed above the earth.

The Gutnish myth runs parallel to Jordanes account in some respects and deviates from it in others. Historical scepticism towards a continuity of an oral traditional preserving the migration of the Goths from Gotland is however justified.  From a linguistic perspective there is also some evidence to substantiate Jordanes’ claim of a Scandinavian origin. The following parallels between North-Germanic and Gothic have led some scholars to argue for Gotonordic prestage (Eckhardt Meinecke 1953: 83-84).

  1. PGmc. –i̯̯i̯-> Goth –ddj- and PGmc. –i̯̯i̯-> OIce. -ggj-

PGmc. –ṷṷ– > Goth –ggw– and PGmc. –ṷṷ– > OIc. -ggv-

  1. retainment of PGmc. *-z# as Gothic -s# and OIc. -r#
  2. retainment of the fourth inchoative class of weak verbs in PGmc. *-nanã
  3. lexical parallels
  4. Loss of the verbs preserved in WGmc. dōn, gān and stān.
  5. retainment of ending of the strong verbs in *-t, e.g. Goth and OIce. gaft against OHG gabi


However striking the parallels, they could also be explained in terms of retaining archaic features than in terms of common innovation. West-Germanic and North-Germanic share more features than an alleged Gothonordic. The inconclusiveness of the linguistic evidence does not give us an argument in favour of the Scandinavian origin myth of Jordanes. Nonetheless,  a linguistic continuity from the 2nd century tribal name Gutones to the fourth and fifth century Gothi and Γότθοι seems probable. Also to be mentioned is the archaism of the Gothic language itself. As a Germanic language attested in fourth century writings it predates the literary monuments of most other Old Germanic languages in atleast three and a half hundred years, so some archaism is to be expected. Most “transformation of the Roman world” historians on the other hand would have the fourth century Goths live in a highly Romanized ethnically diverse cultural realm and considering this historical background the lack of cultural loans from Latin and Greek is perplexing (not counting eclessiatical terminology). This might point to a strong Germanic core to Gothic identity (contra Goffart), an assumption strengthened by the fact that Gothic supplied the neigbouring Alans with quite some lexical material which eventually made it’s way into modern Ossetic, while the other way around the loans seem to be quite limited. Abaev (Moscow 1958) mentions, amongst others, these words:

CGmc. Ossetic Meaning
*aggwus (Goth.) > wyngæg narrow, oppressed, surpressed
*qairnus (Goth.) > koyroj mill
*lǣswaka > læsk pasturage against payment
*nakwina > lægoyn[3] hairless, bald, naked
*gahwi > qæw village, settlement
*rapaina > rævæjnæ long, thick hemp-rope
*wīsa > wis patch with mowed grass
*strab- > sævn width of clothes
*spīra- > fsīr ear of corn
*kurdra- > k’ord group, much, bundle[4]
*spelda- > syvældæg layer

The same assertion holds true for language contact with Slavic. While quite some Germanic words have entered Slavic through (presumably) the Gothic language, the other way around no such borrowing has been active. Consider for example these loans:

OCS PSL Gothic
duma *daumā < dōms
gotoviti *gataṷītēi < gatauiþs < gataujan
kupiti *kaupītēi < *kaupiþs < *kaupjan <  kaupōn
kusiti *kausītēi < kausjan
lěkъ *lēkъ < *lēka < lēkeis
lixva *leixva < leihwa < *leiχa-

An interesting article by Kortlandt (2001)  titled “The origin of the Goths” argues for a Central European urheimat for the ancestors of the Gothic speaking people. He convincingly argues that the Proto-Goths must have been situated more to the west than has often been assumed. The borrowing of the –āreis suffix from Latin –ārius pleads for this, along with the fact that Gothic borrowed the  Latin form of most Greek ecclesiastical terminology. Supposed loans from Celtic also point to a more western origin. Kortlandt does however hold open the assumption that the Gothic identity came from the Baltic Gutones via one of Wolfram’s Traditionskerne. Unfortunately, Kortlandt does not treat the aforementioned loans from Gothic into Alanic and Slavic, but they are easily explained from late fourth century contact when the Goths reached the mouth of the Danube. Especially the Alans remained bound to the Goths for most of the following century, fighting along them at Hadrianople (478), Rome (409) and at the epic battle at Châlons (451) where the romans and their Alanic-Gothic allies went into battle with the Huns of Attila.

This case shows how important an interdisciplary approach is to tackling migration history. One could ask therefore why historians do not make use of linguistic arguments. One reason is probably to be found in the technical nature of the linguistic discipline. Another concerns the scepticism of historians who work in a hermeneutic science in the positivistic approach of the comparative method of linguistics. As historical linguists we should help span the divide between old Germanic philology and medieval studies by making our arguments more accesible to the interested historian; Kortlandt’s article is a good example of this, focussing on cultural loans instead of purely linguistic argumentation.

A thing we could blame the historians for is ignoring the linguistic diversity of late antique and early medieval Europe, restricting themselves to Latinate sources. One need but to remember that Latin was not the only written language in the early medieval west. From the period of 400 – 900 AD we find literary monuments written in Celtic, Germanic and Slavic vernaculars. Considering only a small percentage of the population that lived outside the Romance speaking territory knew Latin, we have to assume that the generational transfer of culture in the Early Middle Ages was mainly done in the vernacular. When one considers the fact that early medieval vernacular sources often deviated immensely from the genres of rigid and devote Latinity, the comparative Indo-Europeanist could only sigh and shake one’s head at how many. Historians should better heed the word of Jordanes himself to “follow the writings of their ancestors and cull from their broad meadows a few flowers to weave a chaplet for those who care to know these things”.


Jay, H. Jasanoff, “Gothic”, in: the ancient languages of Europe, Roger D. Woodward ed. (Cambridge 2008) 189-214.

Augustin Speyer, Germanische Sprachen; ein vergleichender Überblick (Göttingen 2007).

Eckhard Meineke et Judith Schwerdt, Einführung in das Althochdeutsche (Paderborn 2000)

Thomas F.X. Noble, “Introduction; Romans, barbarians and the transformation of the Roman empire”, in: From Roman provinces to medieval kingdoms, Thomas F.X. Noble ed., (New York 2006) 1-28.

Herwig Wolfram, “Gothic history as historical ethnography” in: From Roman provinces to medieval kingdoms, Thomas F.X. Noble ed., (New York 2006) 43-69.

Walter Goffart, “Doest the distant past impinge on the invasion age Germans” in: From Roman provinces to medieval kingdoms, Thomas F.X. Noble ed., (New York 2006) 1-28.

Bryan Ward-Perkins, The fall of Rome and the end of civilization (Oxford 2005).

J. Svennung, Jordanes und Scandia; kritisch-exegetische studien (Uppsala 1967).

“Gutasaga”, in: Altschwedisches Lesebuch, Adolf Noreen ed., (Upsala 1892-94) 37-39.

F. H.H. Kortlandt, “The origin of the Goths”, in: Amsterdamer Beiträge zur älteren Germanistik 55 (Amsterdam 2001) pp. 21-25.

Robert Anderson, “Foreword”, in: The transformation of the Roman World, Leslie Webster et Michelle Brown eds., (London 1997).

Han Nijdam, “Twee aardewerken schaaltjes. Collectief geheugen, (dis)continuïteit in de Friese cultuur, met het fenomeen ‘magie’ als case study”, in: Speculum Frisicum; stúdzjes oanbean oan Philippus H. Breuker, Rolf H. Bremmer Jr. e.a. eds., (Leeuwarden 2001) 59-78.

Vasily Abaev, Istoriko-jatymologičeskij osetinskogo jazyka (Moscow 1968).

[1] My former professor in Utrecht, Mayke de Jong, confessed that she’d rather speak of “Sub-Roman” traditions than of “Germanic” traditions.

[2] [2] Jordanes names is interpretated as *Iƀurnanþs by Grimm who draws for this interpretation on the spelling <iornandes> in some existing manuscripts. He argues that sixteenth century editions always read <iornandes> and may reflect additional old manuscripts that are lost. It is interesting to note that his father is called Alanoviamuth, in which the first element undoubtedly refers to the ethnonym Alani. Maybe the name should be read as *Alano-Weihamōþs “der Kampfmutige der Alanen”. Alanic descent for Jordanes is also suggested by the name of his grandfather, who was named kandag, which could go back to Old-Ossetic *kæn-dag “he who wears sack-cloth”.

[3] Regular dissimmilation of *n…n > l…n as seen in lamaz “Islamic prayer” (< Pers. namāz)

[4] Glossed by Abaev as “группа, множество, стая” in Russian

Before and between Violin and Fiddle

Etymological inquiry into the origin of ModDutch viool and vedel[1] 

Since my native language is Dutch my interest in the etymology of specific words often but not always starts with a simple inquiry into my native idiom. This was also the case when I wondered about the etymology of Modern Dutch viool “violin”, which is of course related to ModE violin and ModFr viole. The direct etymology of the word is to be found in Middle French viole which itself goes back to Old Provencal viola / viula. The source for this OProv. viula is without a doubt Medieval Latin vitula or vidula “stringed instrument, lyre”, but here things get problematic. What is the origin of MLat vitula / vidula? The Latin word is first attested in the eleventh century (MLat vidula) and attempts to connect it with vītulāri “to exult, be joyfull” on the one hand and fidēs / fǐdǐcǔla “stringed instrument” on the other are not without difficulty.

Latin fidēs and its diminutive fǐdǐcǔla are often assumed to originate in a mediterannean substratum language that also yielded Greek σφίδες “tripe fit for cookery”. The Medieval Latin /-t-/ in vitula should then be considered a Romance confusion of voiced and voiceless intervocal obstruents, which seems to be unlikely for a twelfth century attestation. The verb vītulāri seems to be derived from the name of the godess Vitula or Vitellia, who is the godess of joy, so this verb also doesn´t give us a solid etymology for vitula[2].

The modern French word vielle, by most historians of music erroneously thrown at the same etymological heap as ModFr viole and ModE fiddle, is plausibly derived from fǐdǐcǔla, assuming that Romance /e/ < Lat /ǐ/ in this case merged with /ɛ/ that could diphtongize to //. That ModFr vielle does not share the same origin as ModFr viole is clear from the OFr /-ʎ-/ that must come from a palatalized /-kl-/, cf. ModFr abeille “bee” < Lat. APICULA and ModFr caille “quail”< Lat. QUACULA.

When it comes to the origin of Medieval Latin vitula / vidula it seems therefore reasonable to look for that other major source of Medieval Latin idiom, namely the Germanic languages, a solution that was preferred by amongst others Warthburg and Bloch in their Dictionnaire Etymologique du Français. Here we encounter a family of words which is often assumed to have their origin in just that Medieval Latin word vitula, to which the ModDutch word vedel “medieval string instrument” and ModE fiddle belong. This instrument presumably possessed three to five strings and was initially plucked (Johnston 2011: 522). From the 10th c. onwards the bow became current in western Europe (see Woodfield 1984: 9) and the OHG fidula could also be bowed as evidenced by the late OHG gloss viedelstaf “fiddlestick” (Oxford Bodleian Library, Junius 83, f.63r).  The earliest attestation of the ancestor of these Germanic words is to be found in Otfrid of Weissenburg’s Evangelienbuch as fidula.

Otfrid’s Evangelienbuch, V 23, 197-201

Sih thar ouh ál ruarit thaz órgana fuarit
Líra joh fídula joh mánagfaltu suégala
Hárpha joh rótta joh thaz io gúates dohta
Thes mannes múat noh io giwúag thar ist es álles ginuag
There everything is moved By what the organ produces
Lyre and fiddle And many kinds of flutes
Harps and rotes And everything deemed good
which man’s mind always retained There there was enough of that all

The attestation of OHG fidula in the context of this fragment implies that the OHG fidula showed a closer ressemblence to the lira than to the rotta, which were both stringed instruments as well, since the instruments were probably paired for their similar likeness. Otfrid’s Evangelary also argues in favour of regarding OHG fidula as the older form, since it cites OHG fidula next to  OHG swegala “flute”, which precludes confusion of vowels in unstressed syllables. The word fidula  is also attested in a tenth century manuscript (Ro Pal. Lat. 1517) as fidala, glossing fidia for fidicula (confusion with tibia?) in the work of the late antique poet Prudence. However, when one looks at the vernacular glosses it should be noted that fidala and its younger reflexes do not only gloss fidicula, but also tibia “flute” and the nomen agentis fidulāri glosses tibicen “flute-player”. The identification as a string instrument is therefore not certain from the Old High German glosses. In Old English however the identification as fidicula is more certain since OE fiðelere and fiðelestre do gloss Latin fidicen “lyre-player, someone who plays a string instrument”.

The Old Gmc. comparanda can be analyzed in several ways. OE fiðele may point to PE *fiþælu < PGmc. *fiþalō or PE *fiþulu < PGmc. *fiþulō, agreeing with OHG fidula / fidala.[3] OIc. fiðla can go back to PGmc. *fiþlō, but also*fiþulō/*fiþalō. The nature of the suffix variation is as of yet unexplained and has been the main subject of my MA-thesis. To go into it here would not do justice to the complications and nuances of the phenomenon and its explanations and exceed the aim of this post. Suffice it to say that the Gmc. comparanda in the case of our etymon may be explained by PGmc. *fiþlōn .

If one insists on seeking the origin of the Old Germanic words in Medieval Latin vitula / vidula, one is confronted with the very late attestion of vidula (two centuries later than the first OHG attestations!) and, more importantly, the perplexing presence of Gmc /-þ-/ for Latin /-t-/ or /-d-/. My preferred solution concerns onomatopaeic formations which are amongst others to be found in the Romance languages themselves. Here we find quite some words concerning music and sound that have a sound imitative origin, e.g. Provencal piular “to bemoan, to yell”, miular “to cry” and *fiular “to whistle”. The Dizionario Etimologico della lingua Italiana doesnt it hold it unthinkable that the origin of OProv. viola / viula must be sought in a formation originally meaning “lo strumento che va viu” (an instrument that makes a viu-sound). For the plausibility of a sound imitative origin for Gmc *fiþlōn one should also note the etymology of ModGerm geige “violin”, which is the modern reflex of MHG gīge, derived from the verb gīgen “to make the sound gīg”.[4]

In the same way a Germanic sound imitative root *fi- could have acquired an instrumental suffix *-þlōn or, a root PGmc. *fiþ- an instrumental suffix *-n (litterally denoting “an instrument that makes a fi(ð)-sound”), compare Germanic *pīpana (cf. ModDu pijpen / piepen) as a sound imitative verb which gave rise to the nomen agentis *pīp-āri. Something similar was proposed by Van Wijk who insisted on a Germanic formation going back to the PIE root *piH-. This PIE *piH- would also be present in OCS piskati “to pipe” < PSl. *pīskātī, OCS pojǫ, pĕti “to sing” and Toch.B. pīyaṃ []). To this root the PIE instrumental suffix *-tlo/*-tleh2 may have been suffixed, i.e. PIE *píHtlo-/*píHtleh2 >> *fiþla-/*fiþlō with secondary short vocalism, either by laryngeal metathesis or by analogy to the PGmc. *ī/i-ablaut that goes back to PIE *ei/i.. Then OIc. fiðla would reflect the original PIE formation.

Assuming a Germanic origin for the ancestor of ModE fiddle and ModDu vedel, it becomes plausible that Medieval Latin vidula was loaned from Germanic instead of the other way around, the sound substition of */-ð-/ for /-t-/ or /-d-/ being quite common; early /-ð-/ was first romanized as /-t-/, cf. OFrnk. *friþu- > Gallo-Roman fretum, but later developed into Romance /-ð-/ before subsequently disappearing. Naturally, also Germanic /-ð-/ was romanized as /-ð-/ before disappearing, cf. OFrnk. *laðo > OFr. laon “board” and *flaðo > OFr. flaon “specific cake” (cf. ModDu vlade /vlaai).

Concerning the etymology of ModDu viool we could start from Gmc *fiþlōn > OFrnk *fiþula > Gallo-Roman fitula > vidula <viðula> > OProv. viula > MidFrench viole > EModDu viole. The initial Romance /v-/ could have arisen from lenition caused by the article, cf. una  fitula > una viðula, for intervocalic /-f-/ went through Romance /-v-/ as evidenced by OFr. Estievene < STEFANU and ravene < RAFANU. Another solution would be to assume a somewhat later loan from the north of Gaul where dialects of Franconian were still spoken untill around the early ninth century, for Franconian developed /f-/ into /v-/ as evidenced by Franconian gloss material and e.g. the tenth century property list of the bishopsee Utrecht (known in Dutch as het Utrechtse goederenregister), cf. Velepan, Velesan, Vrando. A Franconian loan seems, however, overly complicated and an internal Romance development is to be preferred. The direct Germanic development could be illustrated by Gmc *fiþn > OFrnk. *fiþula > OLFrnk. *vidala > MidDu vedel(e). This in it’s turn would be the reason why we have the pairs fiddle / violin in English and vedel / viool in Dutch.   


Alkire, Ti and Carol Rosen, Romance languages; a historical introduction (Cambridge, 2010).

Beekes, Etymological dictionary of Greek, 2 vols (Leiden 2009).

Bourciez, Edouard, précis historique de phonetique française,  nouvelle collection a l’usage

des classes III (Lille 1921).

Cortelazzo, Manlio et Paolo Zolli, Dizionario Etimologico della lingua Italiana (Bologna 2004).

Gamillscheg, Ernst, Romania germanica. (Berlin:1970).

Greimas, A. J. Dictionnaire de l’ancien français; jusqu’au milieu du XIVe siècle (Larousse, 1989).

Kluge, Friedrich, Nominale Stammbildungslehre der altgermanischen Dialecte (1886).

Vaan de, M., An etymological dictionary of Latin (Leiden 2008).

Zink, Gaston, phonétique historique du français (Paris, 1986).

[1] This article has been edited a few months after it was first published.

[2] the variation vitula and vidula is caused by Romance lenition, compare the Kassel Glosses in giving fidelli in stead of vitelli “calves”.

[3] OHG fidala may also be a later variant of OHG fidula with weaking of old u to a in unstressed syllables.

[4] For the MHG meaning, cp. Strassburger Alexander 210 “sîn meister lartin die seiten zihen, daʒ alle tône dar inne gigen”. That MHG gīgen also meant “the sound that fiddles make” is clearly illustrated in Der Trojaner Krieg by Konrad von Würzburg (13thc. CE) 3211: “daʒ man guote noten gîget ûf alten videlen”.

What did the Carolingians have against the Jews?

The OHG Isidore translation
and it’s background

A friend of mine is doing her PhD on early medieval glossaries, which brings her often to Leiden University to work with facsimiles and original early medieval manuscripts, some of them containing glossaries that are highly valuable to the linguist, although linguists rarely turn to the manuscripts themselves. We had an interesting discussion on why glossaries were collected in the first place and, more importantly, the early medieval origins of vernacular translations on the continent. Somewhere in the discussion I mentioned the Old High German translation of Isidore’s De fide catholica ex veteri et novo testamento contra Iudaeos, remembering that linguistically it was highly archaic which suggested that it must have been written before the ninth century. We both wondered why in heavens name a sixth century tractate agains the Jews was translated in the vernacular in the first place, a wondering which demanded further inquiry.

As far as we both knew the Jews were a quite insignificant minority in Carolingian society (especially in the Germanic speaking parts of the empire) who mostly worked in trade enterprises and held some small but controversial privileges, although being generally discriminated against in legal matters. In this blogpost I want to discuss the main manuscript in which the text is preserved, lay out why the tractate was translated and give you a small fragment of the text, i.e. the latin original, the OHG translation and a modern English translation of the Latin and the Old High German.

                The text that we will discuss is known as De fide catholica ex veteri et novo testamento contra Iudaeos, written by bishop Isidoris Hispalensis, Isidore of Sevilla, around 614-615 as a theological motivation to a royal decree of the Visigothic king Sisebut, ordering all Jews to convert to christianity. The Old High German translation is preserved in a late eighth century manuscript, kept in the Bibliothèque National de Paris (BN lat. 2326), and a fragment of the text is to be found in another manucript, this one preserved in Vienna (ÖNB cod. 3093). Palaeograpically the text in the Paris manuscript, evidently a copy of the original, uses an orthograpical system commonly used in Murbach. The Isidor-fragment is preceded in the Paris manuscript by an OHG translation of  the Matthew gospel, another text which the editors of the manuscript called de vocatione gentium, the final part of an eigth century sermon and sermon LXXVI of Augustine. The Latin and vernacular are given in two seperate columns. Linguistically it seems preferable to place the author of the translation in Lorraine, since the language used is Old South Rhine Franconian. Metz, St. Avold and Hornbach have been suggested as possible places. Kowalski-Fahrun convincingly argued for an early Carolingian date to the translation, since linguistically the Old High German preserved in the texts is quite archaic.

                This can be illustrated by the fact that the common OHG syncope in the past participles and the preterite of the weak verbs has not yet taken place, certain unique OHG forms are used that only have a corresponding form in Gothic (OHG lyuzilla cf. Goth luttila, OHG chillaubin cf. Goth. gilaubi, OHG mittingart cf. Goth midjungarda, OHG geizzsi cf. Goth. gaitin) and certain archaic morphological traits, such as the preservation of the u-declination, the m-ending in the first singular of the present tense, as seen in bim and sculim, and the dative plural ending in –m.

With an early dating of the text corroborated by linguistic arguments, placing the translation in the beginning of the reign of Charlemagne (770’s AD) seems reasonable, although most vernacular writings are connected to late Carolingian scholarship in the closing years of the eighth and beginning of the ninth century.  Even with a late dating of the translation, the OHG Isidore, together with the Monsee-Wiener-Fragmente, would be the oldest witness to Carolingian theological writings in the vernacular.

It is believed that the reasons for the translation of the Latin text in the vernacular were purely practical. For Isidor had collected in his tractate exhaustively all the passages of the Old Testament that had any connection to Christ and he gave an excerpt of all material from the Old Testament, with the exception of the psalms, that was necessary for the support and understanding of the christian faith. It was, as it were, a survey of messianistic prophesies along with the necessary explanations historically ordered. It would have functioned as a practical teaching manual while the original intentions of the sixth century author were pushed into the background. This is corroborated by the fact that Latin iudaeos is translated by unchilaubun, i.e. “infidels” or “unbelievers”, showing that the identification of the antagonists as Jews against whom the tractate was originally meant, was secondary.  The sixth century work contra iudaeos presented to the early Carolingian clergy the quintessence of Old Testamentic writings relevant for a proper understanding of the true faith and it is probably there where we have to look for the reasons behind the OHG translation. Remarkable is the fact that the Latin is translated in what looks like mostly genuine OHG syntax. Also, in a lot of places the OHG is a very loose translation of the Latin, aiming to convey the meaning of the text, not so much the form. The Tatian translation with its word-for-word glossing style is in many ways a step back from the high-level translating skills shown in the OHG Isidore.

                To give you an idea of what the text looks like here a fragment of the OHG Isidore translation, the Latin and the OHG, to which I added an English translation of the Latin and an English translation of the OHG. Personally, I find the late antique and early medieval logic in theological matters that this text conveys highly amusing and risible.

Latin version

4. Si Christus deus non est, dicant Iudaei nobis, quem sit affatus deus in Genesi cum diceret: ‘Faciamus hominem ad imaginem et similitudinem nostram.’ Sic enim subiungitur : ‘Et creavit deus hominem ad imaginem et similitudinem dei creavit illum’. Quaerant ergo quis deus creavit , aut ad cuius dei imaginem condidit hominem quem creavit.

5. Quod si respondeant : ‘ad angelum,’, num angelus aequalem cum deo
habet imaginem, dum multum distet imago creaturae ab eo qui creavit. Aut numquid
angelus cum deo potuit facere hominem ? Quod ita existimare magnae
dementiae est. Cui ergo dicitur, aut ad cuius imaginem conditus homo creditur,
nisi ad eius, cuius una imago cum deo est et unicum nomen divinitatis est.

English translation
of the Latin

4. If Christ isn’t god, as the Jews may say to us, to whom would God have spoken in Genesis when he says: ‘Let us make man in our image and our likeness.’ Because it is joined with the following : ‘And god created man,  in the image and likeness of god he created him’. Thus they may ask which god created or in the image of which god he composed the man that he created.

5. But if they may answer: ‘in the image of the angel,’ does the angel not have an image similar to god, while the image of that being is very different to him who created.  Or surely the angel couldnt make man together with god? To believe this is a sign of great folly. To whom is it said thusly or in the image of whom is man believed to be composed, unless in the image of him, whose one image is with god and is the sole name of divinity?

Eighth century Old Rhine
Franconian rendering of the Latin

4. Ibu Christ got nist, sagheen nu dhea unchilaubun uns, zi huuemu got uuari sprehhendi
in Genesi, dhar ir quhad : ‘duoemes mannan uns anachiliihhan endi in unseru chiliihnissu’. So dhar auh after ist chiquhedan : ‘endi got chiscuof mannan anachiliihhan endi chiliihhan gote chifrumida dhen’. Suohhen dhea nu auur, huuelih got chiscuofi, odho in huuelihhes gotnissu anachiliihhan
mannan chifrumidi, dhen ir chiscuof.

5. Ibu sie antuurdant endi quhedant : ‘in angilo’: Inu ni angil nist anebanchiliih gote? Dhanne so dhrate mihhil undarscheit ist undar dhera chiscafti chiliihnissu endi dhes izs al chiscuof. Odho mahti angil so sama so got mannan chifrumman ? Dhazs so zi chilaubanne mihhil uuootnissa ist. Huuema ist dhiz nu zi quhedanne odho zi huues chiliihnissu uuardh man chiscaffan, nibu zi dhes dher anaebanliih ist gote endi chinamno ist mit godu ?

English translation
of the Old High German

                4. if Christ isn’t god, the unbelievers may say to us, to whom is god speaking in Genesis, where he says: “Let us make man similar to us and in our likeness.’ And where also is said after that:
‘and god created man, similar and liking to god he created him. However they may ask, which god would have created or in the likeness of which god he would have made man, when he created him.

                5. If they may answer and may say: “in the image of the angel’: is the angel not similar to god? Because there is a very big difference in the likeness of that creature and of him who created all. Or would the angel and god have created man together? To believe this is great folly. To whom should this be said or in the likeness of whom is man created, if not of him who is similar to god and who is the namesake of God.


Wilhelm Braune, Althochdeutsche Grammatik (12th edition 1967: Tübingen 1868).

Braune, Wilhelm, Althochdeutsches Lesebuch (17th edition 1994 Tübbingen; 1875).

Eckhardt Meinecke et Judith Schwerdt, Einführung in das Althochdeutsche (Paderborn 2001).

Herta Kowalski-Fahrun, “Alkuin und der Ahd. Isidor”, Beiträge zur Geschichte der deutschen Sprache und Literatur 47 (1923) 312–324.

Horses in the Heroic Age


Johnny Cheung, Studies in the Historical
Development of the Ossetic Vocalism
, Beiträge zur Iranistik 21 (Wiesbaden

W.L. Van Helten,
Altosfriesische Grammatik (Leeuwarden, 1890)

A. Van Loey, Schönfeld’s
Historische Grammatica van het Nederlands
(6th edition; Zutphen, 1959).

Dirk Boutkan and Sjoer Michiel Siebinga, Old
Frisian Etymological Dictionary
(leiden 2005).

a suspicion also uttered by De Vries in his “Etymologisch Woordenboek van het



[4] The wide distribution of similar
forms (e.g. ORus. xъmelь
,  Hung. komlo, Finn. humala) has
been explained by Abaev by positing an Alano-Ossetic origin for all these
words, going back to PIIr. *hauma-aryaka,
the famed haoma of the Gathas, although Cheung sees a formal problem in
that *aryaka should have yielded
Ossetic **ælæg and not ællæg.

Belgian Beer, Belgian toponyms and Belgian i-umlaut

The Belgians make some fine beers…whatever side of the planet you´re
from, it is very likely that you have heard this or experienced this yourself.
If not, go to a pub and get hammered right now! One of these fine beers is
Leffe blond, a soft fruity beer with a bitter aftertase. However, I am not here
to advertise beer brands, I am hear to talk about the name Leffe and its
toponymics. When I was grabbing a beer this week with one of my friends I thought
that the name Leffe, named after the 12th century Belgian abbey
where the beer was originally brewed, contained the toponymical element –effe­,
going back to Pgmc *-aχwjō. Unfortunately, a quick peek in Maurits
Gysseling’s “Toponymisch Woordenboek” proved me wrong by showing the first
attestation of the name as leffle in 1152 AD. This name is probably to
be connected with a proper name Leffo and a toponymical element *-lauha
meaning “woodland”. However, the foundation of the abbey was carried out by
monks from a place called Floreffe, which does contain the element –effe. In
this post I want to explore the specifics concerning the historical development
of this element.

                The toponyms named
by Gysseling as containing this element are the following, given with the
proposed etymology.



First attestation






< Ofrnk *bōn[1]-aχwjō



ca 700

< Ofrnk *flōr-aχwjō




< Ofrnk *hunu-aχwjō




< Ofrnk *leni?-aχwjō




< Ofrnk *marna?[3]-aχwjō



ca 900

< Ofrnk *soni-aχwjō



ca 1070

< Ofrnk *sumbro?[4]-aχwjō



ca 1050

< Ofrnk *walha- aχwjō


The first remarkable thing that springs to mind is that the Old Frankish
fricative /h/ or /χ/ is represented as /f/ in the attestation,
with the exception of the eighth century[5]
attestation florechia for Floreffe. This, however, is a very common
soundchange seen in numerous branches of language families (e.g. Latin, where
PIE */gwh/ turns Italic */χw/ before yielding /f/ in
Sabellian and Latin) and is the regular Romance soundsubstitution of Germanic
*/χ/ in the combinations Old Frankish */χr-/[6]
and */χl-/. Wordinternally, Old Frankish */-χw-/ was replaced
by Pre-French /-ff-/ as shown in the Pre-French latinization of Old Frankish *skōhwāri
as Mlat scoffarius. Wordinternal /-χχj-/ is romanized distinctly as
shown by Old French gehir from Old Frankish *jeχχjan. Consider
the following etymologies.








< *hlanka

MHG lenken


descendant of Clovis

< *hlodowing




< *hlāwa-

Dutch lauw



< *hrok

Dutch rok


to wrinkle

< *hrunkjan

ON hrukka


to hit

< *hrappjan

ON hrappr

gehir (OFr)

to confess

< *jeχχjan

Dutch biecht < *bi-jeχti

scoffarius (Mlat)


< *skōhwāri

Germ Schuster


The development which eventually yielded –effe is best
illustrated by the early attestations of Floreffe, both dating to the eighth
century, although their copies are centuries younger, preserved in eleventh
century manuscripts containing the eighth century Vita Sancti Bertuini.







8th c. 11th
c. Copy

form already underwent
i-mutation, Gmc fricative intact


8th c. 11th
c. Copy

Romance development of Lat
/ō/ > */eu/ reflected in spelling?



Romance substitution of
Germanic */-χw-/ for /-ff-/



first geminate spelling

In the case of Floreffe, the first attestation shows the fricative /χ/
in the spelling /ch/[7],
herein being the only attestation of the toponymical element showing such an
archaic state of affairs. The first vowel of the second element is ostensibly
umlauted under influence of the yod in the following syllable. This umlaut is
evidently of Germanic origin for several reasons. Firstly, Pre-French metaphony
would not have been active in clusters other than /ry/ and /sy/ (e.g.
Vulg.Lat. area > Mod.Fr. aire [ɛr], Vulg.Lat. basiare
> Mod.Fr. baiser [beze]). Secondly, the closed syllable of the
element would have preserved the /a/ if the element was accented
independently (compare Pgmc *þwaχhlja > Mod.Fr. touaille [twajə]).
Finally, a Pre-French soundchange in the second toponymic element would be
highly unlikely since the vowel in the first toponymic element remains
unaffected by Pre-French diphtongization of old /ō/ to */eu/
<e>? (this */eu/ presumably develops to */œu/
before being simplified to /œ/ <eu> of Modern French) in
most of the attestations. Therefore it seems very likely that the first
attestation of the toponym as florechia reflects a development stage
before romanization of the toponym and possibly before romanization of the
place Floreffe itself.

 The fact that it already underwent i-mutation would
point to a early completion of the process since the early eighth century
versions of the abrogans-list still show largely unmutated vowels.  The Germanic umlaut also tells us that the
Germanic language that was spoken in that area was part of the Frankish dialect
continuum, since in other dialects of Old High German the cluster would have
prevented umlaut from taking place. (e.g. compare the usual OHG sahhis,
to the Frankish of the OHG Tatian, sehhis, sehhit), a feature
that Frankish shares with Old English, e.g. OE hliehhan < *χlaχχjan).
 A final remark is warranted by the
historical and modern geminate spelling /-ff-/. Is this a West-Germanic
gemination before yod? It may be that the form Old Frankish *aχwjō already
underwent gemination before yod, thus yielding *aχχwjō, then i-mutation
took place and subsequently the geminate got romanized as [f:]
<ff>, thus *-eχχwja > *-effia. However, MLat scoffarius shows that the cluster /
χw/yielded a geminate anyway.




Edouard, précis historique de phonetique française nouvelle collection a

Gamillscheg, Ernst, Romania germanica. (Berlin:1970).

Gaston Zink, phonétique
historique du français
(Paris, 1986).

Kr. Nyrop, Grammaire
historique de la langue française
, tome deuxième (Copenhagen, 1960).

Ti Alkire and Carol
Rosen, Romance languages; a historical introduction (Cambridge, 2010).

A. J. Greimas, Dictionnaire
de l’ancien français; jusqu’au milieu du XIVe siècle
(Larousse, 1989).

Wilhelm Braune, Althochdeutsche Grammatik (12th edition 1967: Tübingen



[1] Gysseling reconstructs Old
Frankish *bōn in the meaning of “roseau” (reed), however, strictly
speaking also the element *bauna is eligible for consideration, for Old
Frankish /au/ was already monophtongized in the 10th century, as
evidenced in the Wachtendonckse Psalmen.

[2] But hunafie is attestated in
a diploma from 1034, although preserved in a 17th century copy it
clearly shows a more archaic form.

[3] A suffix *-na to Pgmc *maraz

[4] a proper name Sumbro is
suggested by the toponym Sombrin which is first attested as sumbring, showing
a patronymic suffix.

[5] However, the surviving copy date
back to the 11th century.

[6] The Romanisation of Old Frankish */χr-/
however is not unproblematic, where some dialects do substitute /fr-/ and some
/r-/, these differences may reflect a chronological distribution.

[7] a spelling also encountered
in the 6th century Salic law and Merovingian chronicals and

Alanic or Pre-Ossetic glosses in a Byzantine manuscript

Sysse Engberg, while studying Byzantine liturgical manuscripts in the library of the Academy of Sciences in St. Petersburg in 1992, discovered about thirty glosses, written in an obscure language, in the margins of a thirteenth-century Byzantine manuscript preserving an Old Testament Lectionary (προφητολογίον )[1]. Professor Alexander Lubotsky from Leiden University was the first to identify the language in which the majority of the glosses were written as a prestage of Ossetic (henceforth I shall refer to the language as Alanic), a possibility that was first erroneously dismissed by Russian scholars. In 2003 Engberg and Lubotsky published an article[2] with some preliminary findings of their initial research on the glosses. I’d like to present some glosses Lubotsky identified and explained and give my own explanations for two additional glosses that were not yet examined in a scholarly publication. 

The manuscript was copied in 1275 by a deacon called Ioannes for a priest named Chrysos somewhere within the Byzantine influence sphere. On paleographical grounds it is argued that the glossator who introduced the glosses to the manuscript must have lived in the fourteenth or fifteenth century, presumably he was the owner of the manuscript back then. Historically the activity of a Alanic glossator in presumably late medieval Russia is quite significant because the Alans largely disappear from Russian chronicles in the late 13th century [3] (last time mentioned in 1277-1278) only to come in contact with Russia again in early modern times. The glossator was presumably a non-Greek who wrote glosses in his own language in the margins of the manuscript to easily identify the readings for specific liturgical feasts. The manuscript is written in Byzantine Greek and also the Alanic words are transcribed in Greek alphabet. The correspondences between the Greek transcription and the Alanic language which the Glossator intended to render are not unambiguous as can be illustrated as follows.

Greek <α> = Alanic /a/

Greek <ι> = Alanic /æ/

Greek <ε> = Alanic /æj/
(raising of /æ/ in front of following glide?)

Greek <η> = Alanic /æ/ (lengthening of /æ/ in front of final

Greek <η> = Alanic /i/?

Greek <ου> = Alanic /u/

Greek <υ> = Alanic /w/

Greek <π> = Alanic /#b-/

Greek <κ> = Alanic /-g#/

For those of you who are not acquainted with the background of the Alanic language and its successor, modern Ossetic, I will give a short overview. Most IE linguists will be aware of the fact that the modern Ossetic language as spoken in the North Ossetic Republic, an autonomous region in the Russian state, by some 500,000 speakers, continues an array of Iranian dialects spoken on the South Russian steppes for most of antiquity and the early Middle Ages. The speakers of these dialects are to be equated with the Alans of classical and early medieval sources, a confederation of Iranian tribes that played a large part in the Age of Migrations and the defeat of the Huns on the Catalaunian fields in 451. The area where Alanic was spoken was in historical times much larger than where nowadays Ossetic is spoken. Toponymic and hydronymic evidence indicates that the Alanic speaking area extended far westwards. The area where nowadays Ossetic is spoken was probably historically a Nakh speaking area. Quite recent (17th, 18th century) historical migrations moved the Ossetians to their present geographical location.

The following three Alanic glosses have been identified by professor Alexander Lubotsky. In their article Lubotksy draws attention to the fact that some soundlaws which have taken place in modern Ossetic[4] have not yet been active in this stage of the language. The most evident soundlaw is the following:

PIr. /a/ > Ossetic /o/ before –NC#, e.g.

Iron fonɀ < PIr *panča “five” < PIE *penkwe

Iron ɀærond < PIr *ɀaranta “old” < PIE *geronto

This soundchange has been dated to the thirteenth century by Abaev but Lubotsky rightly argues that the attestations in the προφητολογίον together with a fifteenth century Yassic wordlist containing the gloss daban hworz (Digor dæ bonxwarɀ “to you a good day”) prove that the change atleast postdates the Middle Ages.

τῇἁγίᾳμ(ε)γ(ά)λ() β´ πρωΐ (the great and holy Monday)

ἄυτεσήρ στούρ=Alanic *avdisær stur > Digor avdisær (æ)stur “great Monday”

The word Monday is composed of the elements *avd– “seven” (Iron avd– < PIr *hafta < PIE *septm “seven”) and *sær “head” (Iron sær < PIr *sāra- “head” < PIE *krh2os “horn”) literally meaning “the head of seven”. The Alanic word for “great”, stur, also has a solid IE etymology (Iron styr < PIr *stūra– “thick, large”) and is cognate to OI stórr “great”.

Μη(νὶ) σεπτ(εμβρίω) ηἐιςτήνγέννησιντῆςὑπ(εραγίας) θ(εοτό)κου

(September 8, the birth(day) of the most holy mother of God)

ἀστέμακ παν =Alanic *æstæjmag ban > Iron æstæjmag bon “the eighth day”

The interpretation of this gloss is also quite solid. The Alanic word *æstæjmag containes the element *ast– “eight” (Iron ast < PIr *ašta “eight” < PIE *okto) followed by the suffix –æm to make it an ordinal and the suffix –æg to make it an adjective. The Alanic word *ban (Iron bon < * PIr *banú– “light, beam”), meaning “day”, derives from a well established PIE root *bheh2– meaning to “shine”.

Μη(νὶ) νοεμβρίω ιγ’ τοῦ χρ(υσοςτόμου ) (november 13th, Eve of (st.) Chrysostomos)

ζιρήν κὰμ πάν=Alanic *zæræn(æ) kam ban > Iron zærīn kom bon “day of Golden Mouth”

The name of the specific saint that is to revered on november 13th is etymologically translated, since Chrysostomos means “Golden Mouth”. Also in Old Church Slavonic we find the same tendency to translate the name since the saint is called zlatoustъ there, a compositum of zlatъ “gold” and ousta “mouth”.The Alanic “translation” consists of *zæræn (Iron zærīn < PIr *ɀaranya “made of gold” < PIE *gholh3onyo-) and *kam (Iron kom < PIr *kahma- “ palate”?). This last word does not have a secure etymology but comparative data shows that the word must be reconstructed for Proto-Iranian, as evidenced by Persian kām, Pastho kūmai¸ Khotanese kamma, Sogdian k´m. Abaev suggested a pairing with German gaumen and Old English gōma (cf. ModE gums) but since these words must be derivatives to a root *g’heH2u– “to yawn” the connection with the Iranian word is lost. A pairing with the root PIE *k’es– “to cut” might be considered if one takes into account the semantic proximity of words meaning “cut” “gorge” “ravine” ‘ditch” and “mouth” (cf. ModE gorge < Old French gorges “throat, mouth” and German verkehlung the other way around) although a depalatalization of the initial velar consonant has to be assumed.

I took the following glosses from the request for assistence that Engberg published on the web (linguist-list[5]). Unfortunately the words are translitterated from Greek script and not the full Greek sentence that is pertinent to the glossing is given. I have retranslitterated the two words back to what I think was the Greek gloss when one takes the correspondances of Engberg’s translitteration compared to her 2003 article into account. Lubotsky and Engberg are presently working on a publication discussing all the Alanic glosses but since this publication is still forthcoming we have to wait for their expert opinions. I hope that my two identifications will agree with what they have to say on the subject.

(The exaltation of the holy cross) Tzu var urnag : τζουβαρ

Alanic *ɀuar > Iron ɀuar “cross, sanctuary”

The Alanic word for “cross”,*ɀuar,  is easily identified and is probably an early loan from Georgian (Iron ɀwar < Georgian ǰvari “cross” ) The date of borrowing must evidently predate the fourteenth century.

(the day before midpentecoste) 

pi pinlachu tzau pan : χουτζαυ Alanic *xucaw > Digor xucaw “god”

The spaces between the words are not completely consistent with an
isolation of an Alanic word *xucaw, but a quick look on the photograps of
the manuscript that are included in Engberg’s and Lubotsky’s 2003 article shows
that the spacing between the words is quite dependent on the amount of space
left in the margin. Presumed χουτζαυ in my opinion must be identified as
corresponding to Iron
xwycaw, which is also a loan from Georgian. (Iron < * xwycaw Georgian xuci, xucesi “old man, priest”). It is very conceivable
that an Alanic sentence rendering “pentecoste” includes the word for  “god”.

[1] A προφητολογίον is a liturgical book containing readings from the Old Testament to be read during vespers before the Great Liturgical Feasts.

[2] Engberg, S. & Lubotsky, A.M. Alanic marginal notes in a Byzantine manuscript: a preliminary reportNartamongae: the Journal of Alano-Ossetic Studies, II (1-2) (2003), pp. 41-46.

[3] probably due to the Mongol invasions led by Jäbä and Sübä’ätäi in 1222
and the subsequent deportation of part of the Alanic population.

[4] When speaking of modern Ossetic I mean the standard Ossetic language which is based on the Iron dialect. The Digor dialect form which will sometimes be quoted in this article usually is more conservative.

a BIT of evidence for a PIE root “to bite”




The lexical properties of the words "to chew" and "to eat" are closely akin, which can be illustrated by a look at the Romance languages. Modern French manger
and Modern Italian mangiare are reflexes of Proto-Romance MANDUCARE, a word that originally meant “to
chew”, which had replaced the Latin verb
edere in almost all
Romance languages. This lexical replacement was probably motivated by regular
soundchange because medial lenition of /d/ reduced the original Latin word to a
very minimal state. Spanish and Portuguese were only able to retain the Latin
compound verb
comedere which yielded comer in both languages.
However, chance would have it that the Latin verb
manducare is an
interesting word in itself which presents us with some nice linguistic riddles.
So, let’s leave Romance philology on the sideline for the moment and turn
towards some IE problems.


Latin manducare
“to chew” is a secondary formation to the Latin
mandō “to chew, to bite”. The LIV joins Latin mandō to a Vedic injunctive
máthīt  “he robs” and assumes that a
nasal infix into the PIE root *
meth2 “wegreissen” is responsable for the Latin form. De Vaan (1998)
disagrees and proposes that PIE *
mnth2 “to stir” (Vedic mánthati,
Lith. menčiu, męsti) yielded Latin mandō in the same way as PIE *pnth2 yielded Latin pandō. This however prevents Greek μασάομαι and the Greek gloss μάθυιαι in joining in, which is duly accepted by Beekes
(2009). Pairing
μασάομαι to either PIE *meth2 and * menth2 would mean accepting an aberrant soundchange PIE *-th2
Grk –θ-, which isn’t very alluring[1].
However true this may be, a PIE reconstruction for Greek
μασάομαι may lead to another interesting solution. Beekes and
Frisk both consider
μασάομαι a secondary
iterative-intensive formation to a yod-present *mathyō in the same way
as φῡράω is a deverbative formation to φύρω < *phǔryo. This
leads Beekes to reconstruct a Proto-Greek root *mnth– which
he doesnt consider to be of IE ancienity.


I think it is conceivable that the Greek verb goes back to a PIE root *m(e)ndh-,
which would mean “to bite”, and it may be possible to join Latin mandō to
this root as well. De Vaan, following Schrijver, contends that Latin pand- goes
back to Pre-Italic *pndn– which secondarily acquired an a-vocalism
in Latin. Isn’t it possible that Pre-Italic *mndn– got its aberrant a-vocalism
in the same way? If we keep this possibility in mind, we might envision the
developments as follows:


Greek μασάομαι[2] “to chew, to bite” <
[deverbal derivation] Proto-Greek *mathyō < PIE *mndhyoh1


Latin mandō  “to chew” <
Pre-Latin mndnō < PIE *mndhnoh1


OHG mindil ( ON mél and OE mīdl/mīðl
) which is found as a gloss for Latin lupati[3], would, to my
mind, be another candidate for joining this interesting club. OHG mindil goes
back to Pgmc. *menðla- which, in its turn, could go back to PIE *mendhtlo,
since dental clusters preceding a liquid tend to drop the first dental  instead of assibilating (a phenomenon first
described by De Saussure, 1887).


OHG mindil “bit of the bridle” < Proto-Germanic *menðlaz
PIE *mendhtlo


This is, to my mind, a better etymological pairing than the usual
connection to Pgmc. *munþa-, itself a nominalization by accent shift of
a verbal adjective PIE *mntó[4]– (PIE *mńto
> Pgmc. *munþa-). To link OHG mindil to Pgmc. *munþa- a
vrddhi-derivation has to be assumed, which isn’t very appealing.
Furthermore, the original meaning of the formation PIE *mńto must have
meant something like “chin”, as evidenced by Latin mentum and Middle
Welsh mant, which doesnt add up with the fact that also the tlo-derivation
must be of PIE age, making it a instrumentalization of the noun “chin”. The
fact hat the tlo-suffix makes nomina instrumenti from verbs, not from
nouns, makes this development problematic as well, although this argument may
be countered by assuming that we are dealing with an instrumental lo-suffix[5].
 For these reasons it isn’t very
plausible that Pgmc. *menðla- belongs to Pgmc. *munþa-. The
contention that also Middle Irish métal “belly” belongs to the group of
Latin mentum, Pgmc. *munþa- is semantically and phonologically
not compelling and not plausible.


A last argument may be that a “bit of bridle” is within Germanic
lexically associated with the verb “to bite”, as evidenced by the causative
verb Pgmc. *baitjanan and the noun Pgmc. *bitō. A
derivation from a noun meaning “mouth” is not paralleled within Germanic. Also
in the early Romance realm a derivation from a verb refering to what the animal
does with the bit is preferred, as evidenced by Old French gobel <
Gaul. *gobbo < PIE *g(h)obh– “to
swallow” (cfr. Russian zobát’ “to masticate”)[6].


By positing a PIE root *m(e)ndh-, I think we have
ourselves an acceptable IE etymology for a verb which Beekes considered to be
Pre-Greek. Furthermore, we can assume that Latin and Greek both retained the
original meaning of the root. And, last but not least, we have a better
etymology and understanding of the development of OHG mindil and its



Eugen Hill, Untersuchungen zum inneren Sandhi des Indogermanischen; Der
Zusammenstoss von Dentalplosiven im Indoiranischen, Germanischen, Italischen
und Keltischen
, Münchner Forschungen zur historischen Sprachwissenschaft
band 1, Peter Schrijver et Peter-Arnold Mumm eds., (Bremen 2003).

M. de Vaan, An
etymological dictionary of Latin
(Leiden 2008).

Beekes, Etymological dictionary of Greek, 2 vols (Leiden 2009).

Helmut Rix, Lexikon der Indogermanischen Verben; Die Wurzeln und ihre
(Wiesbaden 2001).

[1] Also in the case of Greek οἶσθα it isn’t very
probable that the laryngeal had something to do with the aspiration (Beekes

[2]  (Greek μασάομαι  <
Proto-Greek *mathyāyomai < QIE *mndh-yeh2-yo-mh2(i))

[3] Traditionally OHG mindil is
translated as “bit of a bridle”.

[4] The verbal root is often
interpreted as PIE *men “hervorragen”, which is also found in Latin promineo
and mons, montis

[5] To my mind, PIE –lo-, -tlo- an
-dhlo- are allomorphs of eachother, a contention that I will make in my thesis.
However, I cannot expand on it now.

[6] A counterargument may be found
in Slavic, where the word *ūzda and its reflexes are derived from the
word for mouth. However, most Slavic languages point to the meaning “bridle”
and only in Čakavian it developed the meaning “bit of a bridle”.    


Illegal clusters in Latin and proto-Romance and what they have to do with an Old High German problem

In my last blogpost I talked about sound laws which cross language boundaries and more specifically the early medieval language boundary between the Germanic and the Romance speech communities. In that respect I want to adduce another case which in my opinion may constitute another example of bilingual communities perpetuating a regular sound change from the one language as a sporadic sound change in the other.  In this specific instance I want to talk about a sound change which has proven to be remarkably persistent, namely the Latin soundlaw which turns /-tl-/ > /-kl-/. This soundlaw is best illustrated by the Latin reflex of the PIE instrumental suffix*-tlo/-tleh2 which has the unmarked Latin form /-culu-/ and /-cula-/.


PIE *peh3tlom > pre-Latin *pōklom >Latin pōculum

PIE *eǵhtlom > pre-Latin *veheklom >Latin vehiculum


Remarkably this soundlaw was also active in proto-Romance and turned the new */-tl-/ clusters, caused by early Romance syncope, into */-kl-./


Vulgar Latin VETULA /wɛtula/ >  *vɛtla > *vɛkla

 > *vɛkja > Italian /vɛk:ja/ <vecchia>.

                > Old French vjɛλə > French /vjɛjə/ <vieille>

 The pre-stage *vetla probably never was a phonetic reality because /-tl-/ constituted an illegal cluster in proto-Romance which was automatically substituted by /-kl-/. It is interesting to note that /tl/ is still an illegal cluster in present French when it appears initially, as evidenced by a 1998 study of Hallé. In a phonetic experiment speakers of French were asked to transcribe four nonwords containing the initial cluster /tl/ and four nonwords containing the medial cluster /dl/. More than 80 % of the participants transcribed the cluster /tl/ as /kl/, for example in the case of the first nonword.


/tlabdo/ : <klabdo>

Word medially the cluster /tl/ in modern French gained a secondary schwa as is the case with the word <atlas> /atəlas/. In that case it uses a different strategy to render the illegal cluster /tl/ pronouncable than its proto-Romance predecessor.


The tlo-suffix was also productive in the Germanic languages, where it participated in some complicated internal sandhi developments (which I will try to elucidate in my MA-thesis). One of the outcomes of PIE *-tlo is proto-Germanic *-þla. An interesting problem concerning this þla-suffix is the alternation /þl/ to / χl/ which seems to be responsable for OHG mahal in constrast to Gothic maþls. Also OHG bihal, which is thought to derive from *bīþl < *bhiHtlo, shows the same alternation. What happened here? Most Oldgermanicists regard it as an isolated soundlaw in west-Germanic, which has no clear parallel inside the continuum of Germanic languages. This, however, is not a very satisfactory explanation, because only Old High German and Langobardic, which may be regarded as an archaic dialect inside the Old High German continuum, appear to show the variant with /χ/ followed by an anaptyxic vowel /a/ consistently.


If one considers the proximity of the Old High German dialect continuum to the Romance continuum, especially pre-Italian Romance, an interesting possibility comes to mind. A Romance intermediate stage, with the substitution of /tl/ to /kl/, in the development of Old High German mahal from Germanic *maþlaz may yield a solution which accounts for the phonological intracies. Let’s assume, for my hypothesis’ sake, that the word reached early medieval Italy as Germanic *maþl(s). Seventh-century Langobardic /þ/ is consistently substited for /d/ by pre-Italian Romance speakers, as is seen in Italian:


Italian guadagnare < Langobardic *waiðanjan


Therefore I do not think that Germanic *maþl(s) entered pre-Italian Romance via Langobardic. Loans from Gothic into pre-Italian Romance, on the other hand,  substitute /þ/ word medially for Romance /t/.


North-Italian grinta < Gothic *grimmiþa


I’d like to hypothesize that Ostrogothic maþls entered pre-Italian Romance in the very early sixth century and because Gothic /þ/ was word medially perceived as /t/ the Romance speakers would have approximated the Gothic pronunciation of the word as *matl(s). This however was an illegal cluster in  Romance and was perceived as *makl. The word, audibly Germanic in origin, was picked up by the Langobardic invaders of Italy in the late sixth century. Here the anaptyxis kicks in and gives *makal. The thing that happens next is of course the Old High German sound shift which turns medial /k/ into /χ:/, spelled as single <h>or double <hh>, and that would yield the desired outcome /maχal/ <mahal>, which is found in the Langobardic laws.



Gothic /maþls/ > (Italian Romance */matl/ ) > Italian Romance*/makl/ > pre-Langobardic */makal/ > Langobardic /maχal/ <mahal>


This scenario is to my mind more plausible than to postulate an isolated Germanic soundlaw with no phonetic parallels within the Germanic dialect continuum. The form /maχal/ subsequently spread across the Old High German continuum and reached Saxony in the ninth century. Old Saxon mahal is to my mind a loan<!–[if !supportFootnotes]–>[1]<!–[endif]–> from Old High German which isn’t very surprising if one considers the fact that the Franks institutionalized the mahal as a tool of government.



Wilhelm Bruckner, Charakteristik der germanischen Elemente im Italienischen (Basel 1898).


Benedicte Nielsen, “On Latin instrument-nouns in */-lo/”, in: Indo-European word formation; proceedings of the Conference held at the Universisty of Copenhagen, October 20th -22 nd 2000, eds. James Clackson and Birgit Anette Olsen (Copenhagen 2004) 189-213.


Pierre A. Hallé e.a., “Processing of Illegal Consonant Clusters; a Case of Perceptual Assimilation?”, Journal of Experimental Psychology; human perception and performance vol 24, no 2 (1998) 592-608.


Ti Alkire and Carol Rosen, Romance Languages; a historical introduction (Cambridge 2010).

<!–[if !supportFootnotes]–>


<!–[if !supportFootnotes]–>[1]<!–[endif]–> The native Old Saxon term is fortunately also attested as madal.


Crossing the language border

a connection between Old English and Old French?

Every student of Germanic philology and historical linguistics will be familiar with the fenomenon of breaking. This means I don’t have to go in too much detail. To make it short; a select group of Old Germanic languages has the tendency to diphtongize short front vowels to falling diphtongs under influence of following consonant clusters or non-front vowels in the next syllable. The languages in question who show this phenomenon are Old Norse, Old English and Old Frisian. Although all these language may do this as a meance of distance assimilation, the conditions which are needed for breaking to take effect differ among the languages.  The kind of breaking which concerns us here is found in Old English and concerns the breaking of Pre-OE */æ/ > /ea/ under the influence of an immediately following consonant cluster consisting of a liquid and another consonant. Because the resulting diphtongs partook in the Pre-OE phenomenon of i-umlaut we know that OE breaking must predate i-umlaut alltogether. Another thing that is quite clear is that the degree in which the Old English dialects underwent breaking differs significantly. In the case of breaking of */æ/ when followed by a liquid and another consonant the difference mainly comes down to an opposition between the Anglian dialects and the Saxon dialects, the latter one joined by the Kentish dialect. The northern dialects show retraction of Pre-OE */ælC/ > /alC/ in contrast to the West-Saxon development Pre-OE */ælC/ > /ealC/.

Pgmc *haldanan > Proto-West-Germanic *haldan > Pre-OE *hældan > Anglian haldan

Pgmc *haldanan > Proto-West-Germanic *haldan > Pre-OE *hældan > West-Saxon healdan

On the whole, the southern dialects of Old English seem to like breaking better than their northern relatives[1] and West-Saxon and Kentish exhibit the most instances of breaking. Phonetically we may assume that the breaking of the vowel first yielded a diphtong /æa/ and subsequently stressed the contrast by turning it into /ɛa/, orthograpically represented by <ea>. This becomes relevant when we consider a similar phenomenon in old French.

What’s the deal with Old French, you might ask. Well, Old French is amongst the Romance languages quite the odd man out, which is probably due to the heavy influence of second language acquisition by Germanic speakers. This influence was the largest on the lexicon but also greatly affected the morphoponology and syntax of the pre-French Romance language. This is not very surprising if one takes the high degree of diglossia and bilinguality in the Gaulish realm into account.[2] This created a situation in which isoglosses and linguistic tendencies easily moved between the languages spoken by the specific bilingual and diglossal community. For example, the Old High German diphtongization of /ō/ > /uo/ probably originated in the continuum of Romance dialects who all diphtongated Vulgar Latin /ɔ/ to /uo/ or /ue/. Subsequently  this diphtongization wave reached the Germanic speaking realm via Pre-French.

The Old French phenomenon I’d like to discuss with you concerns the Vulgar Latin vowel */ɛ/ which is the reflex of Classical Latin /ě/, this in contrast to Vulgar Latin */e/ which continues Classical latin /ĭ/ and /ē/. In Old French Vulgar Latin /ɛ/ is one of the vowels that are seemingly affected by the vocalization of the Romance velar /l/. The effects of this velar /l/ are visible in the orthography of modern French, because it has remained notoriously conservative.

Vulgar Latin ALBA “white, dawn” > Pre-French /albə/ > Old French (XII century) /awbə/ > Modern French /obə/ <aube>

Vulgar Latin SOLIDARE “solder” > Old French /solder/ > Early Modern French (XVI century) /suder/ <souder>

If one takes Pre-French syncope into account, we can see the development to consist of Pre-French */alC/ > Old French /awC/ and pre-French */olC/ to early modern French /owC/. However, one of the the purported effects of velar /l/ on a preceding vowel which stand out amongst the other effects concerns Vulgar latin */ɛ/. What seems to have happened is that somehow the velar /l/ in combination with the preceding */ɛ/ yielded an Old French triphtong /ɛaw/ or /eaw/ with a dialectal variant in /jaw/ as evidenced by the Modern French spelling <eau> and the Old French spelling <iau>. Modern French has some words continuing the triphtong <iau> which are mainly from Picardian origin, fabiau < *faběllus and depiauter < piau < *pellis. But the usual Old French triphtong is <eau>:

Vulgar Latin BELLITATE > modern French beauté

Vulgar Latin BELLUS > modern French beau

Frankish *helma > Vulgar latin HELMU[3] > modern French heaume

The question which should be asked, if this development is to be equated with the other effects of velar /l/ on vowels, is why only the development of Vulgar Latin */ɛ/ is so aberrant. Why was it the only one to yield a triphtong? If it was indeed, as insinuated by Ti Alkire and Carol Rosen (2010) and explicitly stated by Gaston Zink (1986), a bridging vowel between the front element and the back element of a former diphtong, why did pre-French /iwC/ and /ewC/ not triphtongize to /ɛaw/ or /jaw/? Following Edouard Bourciez, I deem it more likely that /ɛlC/ first broke to /ɛalC/ and only after the breaking the liquid vocalized to /w/. So the development of the example given above could be imagined as follows:

Vulgar Latin BELLITATE > [Late Vulgar Latin] /bɛltate/ > */bɛltat/ > [OldFrench] /bɛlte/ > */bɛalte/ >  /bɛawte/ > [modern French] /bote/ <beauté>

Empirical proof for this the /ɛalC/ stage in exactly the form hypothesized is attested in the Bisclavret by Marie de France.

beals chevaliers e bons esteit                       a handsome and good knight he was                       

e noblement se cunteneit                            and he conducted himself in a noble way

The attestation of the precise form <ealC> continuing Vulgar Latin */ɛlC/ strengthens my hypothesis. Unfortunately, I don’t have the time to embark on a quest for further nice orthographic representations of this pre-stage, but the example of /beals/ cited above makes it clear that the analysis of the phenomenon by Gaston Zink has become highly tenuous for it doesn’t seem to represent the empirical data.

We can conclude from the text fragment cited above that the development of */ɛlC/ to eventually the triphtong <ɛaw> takes place fairly late. Some other early Old French texts from the late eleventh and early twelfth century still spell <el>, as evidenced by the “chancun de Guillelme”. In in its title it still spells /Guillelme/ instead of later Old French /Guillaume/ (the glide in the diphtong merged with the palatality of /ʎ/ spelled <l> ). In the case of the development of BELLITATE to beauté, we also find a nice example of <el> spelling in this poem:

Il at perdu sun noble barné                             he has lost his noble knighthood

De dulce france la flur e la belté                    the flower and the beauty of sweet Francia

Gaston Zink explains the occurence of the Old French form belté as indicative of the fact that only stressed Vulgar Latin */ɛ/ yielded a diphtong and considers Modern French beauté as a derivate from the adjective beau. However, the protagonist of the poem itself is called guillelme, a perfect continuation of the Germanic name *wilhelm. When we assume that the Germanic word retained its original stress on the first syllable we have another instance of atonal */ɛlC/ which eventually triphtongizes to / ɛawC/. The proper name also shows that the vocalization of velar /l/ probably wasn’t the first step in the triphtongization of */ɛ/, otherwise we would expect to find guilleume in a text from this period.

What strikes me is the fact that these phenomena in Old French and Old English seem to have developed in a parallel fashion, both showing a breaking of a low front vowel to /ɛa/ before a following cluster consisting of an /l/ and another consonant. I think it may be possible that these phenomena aren’t only typologically related but that the one phenomenon might have caused the other due to language contact. It should be remarked that the fact that only Vulgar Latin */ɛ/ breaks in front of a liquid and another consonant neatly lines up with the fact that the Saxon dialects of Old English only consistently break Pre-OE */æ/[4] to /ea/ under the same condition. The difference between /æ/ and /ɛ/ on the vowel triangle is minimal and the allomorphy in the different realms of Anglosaxon morphology may have driven bilingual Anglo-Normans to transfer the Anglosaxon sound change to Old French.  In that the regard, the date at which this development took place in Old French seems suspicious to me, immediately following the onset of Anglo-Norman relations. The English channel,which nowadays separates the Anglophone nation from the Francophone nation, constituted in the early Middle Ages a vital maritime bridge between the two speech communities, for they were ruled by the same elite.

This, however, brings us in the realm of historical sociolinguistics and in that respect the hypothesis seems moribund. The Anglo-Norman ruling elite almost certainly favoured the use of Old French and it isn’t very likely that they acquired such a good command of Anglosaxon to allow synchronig sound changes to jump the language border. Finally, the similarity of the phenomena between Old English and Old French is rarely (to my knowledge) noted and is indicative of the necessity for the historical linguist to look beyond the linguistic borders of one’s own specialization. The fact that for most Old Germanicists the publications in French, Spanish and Italian aren’t readily accesible without an academic translation doesn’t improve the situation. And unfortunately, most Indo-Europeanists, who do tend to read a lot of modern languages, aren’t that interested in Romance linguistics, because not much information concerning the Indo-European proto-language can be mined from its historical development. I want to end this article with the contention that the development of the triphtong /ɛau/ in Old French is preceded by a stage of breaking and we therefore may very well speak of Old French breaking, independent of the fact what caused it.  


Ti Alkire and Carol Rosen, Romance languages; a historical introduction (Cambridge, 2010).

A. J. Greimas, Dictionnaire de l’ancien français; jusqu’au milieu du XIVe siècle (Larousse, 1989).

Gaston Zink, phonétique historique du français (Paris, 1986).

Kr. Nyrop, Grammaire historique de la langue française, tome deuxième (Copenhagen, 1960).

E. Einhorn, Old French ; a concise handbook (Cambridge, 1974).

R. Girvan, Angelsaksisch handboek, Oudgermaanse handboeken IV (Haarlem 1931).

Bourciez, Edouard, précis historique de phonetique française,  nouvelle collection a l’usage

des classes III (Lille 1921).

Augustin Speyer, Germanische Sprachen ; ein vergleichender Überblick (Göttingen 2007).

Joseph Wright and Elizabeth M. Wright, Old English Grammar, the sudent’s series of historical and comparative grammars (Oxford 1914).

A. Campbell, Old English Grammar (Oxford, 1959).

Gerhard Rolfs, Vom Vulgärlatein zum Altfranzösischen; Einführung in das studium der altfranzösischen Sprache (Tübingen 1968).

[1] The usual caveat concerning Old English dialectology is in place here, pointing at the fact that OE dialectology is mainly dependent on codological and palaeographical clues to the provenance of specific manuscripts and therefore many assumptions on the phonological representation of graphemes are clouded by different copyists and manuscript traditions.

[2] Although many Oldgermanicists and even Indo-Europeanists tend to shun the Romance languages, the developments in these languages are vital to understanding the linguistic tendencies which were common in the linguistic area of western Europe during the early middle ages. The early middle ages are linguistically characterized by a high degree of diglossia concerning Latin and the vernacular and a high degree of multilinguality concerning the different vernaculars spoken in the different areas of western Europe. Especially in the later days of the migration age, when many nations roamed the former provinces of the Roman empire, not only variants of Germanic, but also of Slavic, Avar and Alan could be heard in the encampments of the barbarian armies.

[3] The fact that Romance probably first latinized the Germanic word for “helmet” before accepting it into their lexicon is evidenced by Old Spanish yelmo.

[4] I am aware of the fact that West-Saxon and Kentish also seem to break /e/ > /eo/ before /lh/, but this breaking isn’t consistently attested in our sources and the conditions for this breaking would scarcely be noted by speakers of Old French because the fricative h /x/ wasnt part of their phonology. Also, if some attempt by Old French speakers was made to break their native /e/ to /eo/ this would be clouded by the subsequent vocalization of the /l/ >/w/ because the two back elements would certainly have coalesced.