Bad Romance: An Introduction to the Appendix Probi

Reading Time: 10 minutes

How exactly did we get from Latin to the many Romance languages of today? What changes happened to Latin that it should end up looking so different?

These are the driving questions of Romance historical linguistics, the field of study that aims to investigate how French, Spanish, Catalan, Italian, Romanian and many other languages developed over time from a common source.

In this great endeavour, historical sources are of paramount importance. While the Romans have left us a wealth of writing in their pretty standardised Latin, sources that give us glimpses through the veil of the standard language are rarer and very precious. Latin kept on changing as the first centuries of the first millennium AD went by. Many of these changes were noticed; Latin was a living language, and so we can expect it to have had its fair share of conservative speakers, who tried to rein in any innovation and hold Latin to its classical standards.

This blog post is all about an awesome historical document that captures both the ongoing and inevitable changes within Latin, and something of the negative attitudes towards them. That document is the Appendix Probi.

Now, this has nothing to do with any part of anyone’s intestines. The Appendix Probi (the ‘supplement to Probus’) is partly nothing other than an ancient list. It’s therefore a very accessible text, repetitive and without any complicated grammar, yet its simplicity masks its enormous value for historical linguistics. It is, in my humble opinion, extremely cool.

You can read the full list online here. However, it’s just the plain text. So, here to accompany it is my introduction to the Appendix – the what, the when, the who and the why you should care.

What exactly is the Appendix Probi?

Let’s start with the basics. Strictly speaking, the Appendix Probi is a collection of five medieval, handwritten texts. Their language is Latin. They are written on parchment, have the official designation MS Lat. 1, and are housed in the Biblioteca Nazionale in Naples, Italy. They have the name Appendix Probi because they are attached to a copy of the Instituta Artium, a work (wrongly) attributed to Marcus Valerius Probus, a Roman scholar of the first century AD.

The third is the most famous of the set, so renowned that the title “Appendix Probi” is often used to refer to the third text specifically. This text is just a list. It contains 227 entries, each one with the same format: X, non Y. This, not that. This, not that. For example:

  • 29. auus non aus
  • 37. equs non ecus
  • 64. senatus non sinatus

It’s a list of spelling mistakes and corrections for them! Specifically, it was most likely composed as a guide for scribes writing in Latin.

Unfortunately, images of the manuscript available online are hard to come by, and none of those I’ve found are of good quality or in the public domain. The black-and-white image below (from here) gives us a snippet of what it looks like. For each entry, you can see the Ñ that stands for Latin non.

From here.

The thing is, we don’t tend to write laws and rules that condemn things that people are not doing. In other words, the ‘mistakes’ bear witness to what people at the time might write instead of the ‘correct’ versions. Since there must have been phonetic reasons for why people were making these mistakes, the Appendix Probi must therefore also bear witness to what they were saying too.

What’s very cool is that these mistakes and new forms endured. In them, we can spot many features of sound and grammar that are now prevalent across the Romance family. The author of the Appendix may have condemned them, but they were the seeds of languages to come.

When was the Appendix Probi written?

As with all ancient texts, the provenance of the Appendix has many steps. It survives in one single copy that can be traced back to its production at the monastery of St. Columbanus in Bobbio, northern Italy, sometime between the 7th and 8th centuries AD. The consensus though is that it was not an original creation of the monks of Bobbio, and that the content is in fact a couple of centuries even older.

Bobbio Abbey today. Image from here.

This proposal of an early origin draws support from various sources. For one thing, the text includes errors, like repetitions, which look like the mistakes of a copyist. For another, there is the contextual and linguistic evidence of the words themselves. The inclusion of words like parentalia, the traditional Roman festival of family ancestors, suggests a cultural context that is still partly pagan. Then again, it also includes the spelling for the word septizonium, a likely reference to a grand building in Rome whose construction was ordered by the emperor Septimius Severus. If so, then the Appendix cannot predate the emperor, the building and its dedication in 203 AD.

All of this evidence has been used by scholars to propose a range of dates for the creation of the Appendix Probi. Ronald Quirk, perhaps the world’s leading scholar on it today, suggests “a time-frame between the third and fifth centuries AD” (2017: 1). This places it linguistically within the Late Latin period, and so its spelling mistakes reflect the Latin of that time.

Who wrote the Appendix Probi?

Nothing about its author(s) and the original uses for the text is certain. It is an anonymous document that just happens to have placed in a codex along with the Instituta Artium. The two need not have always gone together. Even the connection back to the grammarian Marcus Valerius Probus is unbelievable, since he lived in the 1st century AD, yet the Instituta Artium mentions the Baths of Diocletian, an emperor of the 3rd and 4th centuries.

Historical speculation can take us somewhat further though. The author was likely a scholar, or at least someone who worked with the creation of texts, and had a good knowledge of Latin and Greek writings. He (or she, but most likely he) may also have had an educational role, since we can presume that the text served as a usable guide for correcting others.

Second-century AD relief of a Roman teacher and three pupils, found in Neumagen, Germany. Image from here.

Why does the text matter?

The Appendix Probi gets referenced a lot. I’m very used to it appearing at least once in academic books on Latin, the Romance languages and language change. As F. J. Barnett puts it:

“[It] has probably been more widely referred to, and quoted from, by Late Latinists and Romance philologists than any other ‘Vulgar Latin text’ – a popularity doubtless favoured by its convenient brevity and the fact that all its entries ostensibly attest vulgarisms, whereas other such texts present only a smattering of them.”

Barnett 2007: 2

Its significance lies in the fact that the words of the Appendix display many features that are different from typical Classical Latin, yet recognisable from Romance languages. It is for some their earliest appearance in the historical record. This extraordinary survival is therefore of vital importance in understanding the process of how and when one language became several new ones. Without texts like it, the many steps in that process would be much more mysterious.

Without further ado, let’s dive into eleven examples of those features.

3. speculum non speclum

speculum: mirror

In the condemned spelling, speclum, we can see the effects of the process of syncope. This is when a sound inside a word is lost over time. It happens in particularly with short, unstressed vowels, like the -i- in English family or the -u- here in speculum. The form with the syncope, speclum, went on to provide Spanish and Italian with their words for ‘mirror’, espejo and specchio.

First-century AD fresco of a woman using a mirror. Image from here.

5. vetulus non veclus

vetulus: old

Veclus likewise shows the effects of syncope, but also a swapping of sounds. After the middle vowel dropped out, this left Latin with an awkward /tl/ sequence of sounds. This was clearly exchanged for a more comfortable /kl/ cluster. It’s specifically veclus that has led to French vieux, Italian vecchio and Spanish viejo, all meaning ‘old’. Through veclus, we can appreciate their connection to borrowed words in English like veteran.

53. calida non calda

calidus: hot

Here we have syncope again, and again, it’s the syncopated form calda that has won out today. Compare French chaude and Italian calda.

56. tristis non tristus

tristis: sad

This one shows a grammatical change. Classical Latin had a series of declensions, by which nouns and adjectives like tristis were grouped together according to certain common features. Tristis belongs to the third declension, but here it’s been given the -us ending of the second. We may take this as evidence of the general reduction (or streamlining) of the old declensions. Interestingly, Italian today has both forms; tristus has become the literary tristo ‘wicked’, while the more common triste ‘sad’ is a reborrowing of Latin tristis.

139. aper non aprus

aper: boar

Here again, the ending of the preferred classical form (aper) has been made regular, giving it the common masculine ending -us.

170. socrus non socra

socrus: mother-in-law

This also involves a bit of grammatical rejigging. Socrus means ‘mother-in-law’ and unsurprisingly is feminine in its grammatical gender. Yet the -us ending had evidently become too closely associated with the masculine gender and male things, so it has been swapped for the more safely feminine ending -a. The new form has endured until today as Italian suocera, Catalan sogra and Spanish suegra.

171. neptis non nepticla

neptis: granddaughter

Nepticla displays another preference in Late Latin and Romance: using diminutive forms of words, instead of the basic word that they are built on. Diminutives often oust the basic form because they have connotations of casual language, and they can also be more regular in their grammar.

We can also see this in the line auris non oricla (#83). While auris is the basic Latin word for ‘ear’, it’s the Latin diminutive oricla that is the specific origin of French oreille and Italian orecchio.

189. bipennis non bipinnis

bipennis: two-winged

The change in vowel here likely reflects an important merger in the vowel system of Late Latin. The short vowel represented by the letter I merged with the long E vowel, leading to a lot of confusion in their spelling in the Appendix. Through this change, for instance, Latin pirum ‘pear’ has become Italian pera. I’ve mentioned this and other Italian changes previously in this post.

A memorial spotted (by me) in the Zadar Archaeological Museum, which mentions not the goddess Diana, but Deana.

207. hostiae non ostiae

hostia: sacrifice, victim

This shows a famous difference between Latin and its descendants: the loss of the consonant /h/ (as in English hat). Its absence is well documented in many ancient sources, and it can be taken as a sign of sub-elite language. It is still absent from the sounds of Italian today and also of French, although French would later gain the sound from the Germanic-speaking Franks, before losing it again. Fun fact.

221. vobiscum non voscum

vobiscum: with you

Here we see another grammatical shift. Vobis and vos are two forms of the same pronoun in Latin, differing only in their case. Where we would expect vobis according to the rules of the classical grammar we learn in the classroom, we instead find vos. This may result from the ongoing breakdown of the Latin case system. This was a slow process, but has today resulted in the absence of case for nouns in most Romance languages.

224. olim non oli

olim: at that time

Finally, this spelling correction gives us further evidence for the decay of nasal sounds at the end of words. This appears to have been going on for some time. What we once full vowels and consonants may have become nasal vowels, which meant that they endured in writing. By this time though, the general confusion in spelling suggests that the nasal sounds had disappeared altogether. We even see them being written where they shouldn’t be, such as in Hercules non Herculens (#19). This loss was a factor in the decline of the case system; it meant that Latin amorem ‘love’ (accusative case) and amore (ablative case) ended up sounding the same, and it is simply amore that exists in Italian today.

So, to conclude

Simply put, the Appendix Probi is a rare gem. Its unknown author has bequeathed an invaluable resource for us historical linguists. We could infer many of the changes that happened in Latin on the basis of Classical Latin and later Romance alone, but having actual evidence for them strengthens and improves our understanding.

What I like in particular is how reasonable many of these changes are. Language is always in a constant process of self-regularising, reducing complexities and irregularities in favour of more productive and transparent systems (while at the same time accidentally creating what will become new irregularities). For instance, Latin has several declensions for its nouns and adjectives, and grammatical gender does not neatly match up to them. What we see in the words of the Appendix is the streamlining of declension and gender. An arguably more efficient system is emerging, built around the masculine ending -us and feminine -a. This system of gender continues to be widespread across the Romance languages today.

Much more could be said (the list is 227 entries long!), but I hope this all serves as a good introduction. So much can be gained from interrogating these old texts. If you know a Romance language, why not give the Appendix Probi a read? You might recognise some things. It’s treasure trove of lost language, and it continues to spoil us today.



  • Barnett, F. J. (2007). The Sources of the “Appendix Probi”: A New Approach. The Classical Quarterly 57(2). 701-736. Cambridge University Press.
  • Quirk, R. J. (2005). The “Appendix Probi” as a Compendium of Popular Latin: Description and Bibliography. The Classical World 98(4). 397-409. Johns Hopkins University Press.
  • Quirk, R. J. (2006). The Appendix Probi: A Scholar’s Guide to Text and Context. University of California.
  • Quirk, R. J. (2017). Hypercorrection in the Appendix Probi. Philologus, 161(2). 350-353. De Gruyter.

2 thoughts on “Bad Romance: An Introduction to the Appendix Probi

  1. Enjoyed the many examples and what we can learn from them. Fun title, that the list author wanted to slap the emerging Romance languages on the nose but didn’t succeed in resisting them:)

    Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: