History of the Latin Language and the Formation of Romance Languages

The Origins and Expansion of Latin

Latin is the language of ancient Rome, with its origins traceable to the $8^{th}$ century a.C. Historically, the legendary date of the foundation of Rome is cited as $753$ a.C. Initially, the language was spoken only in Rome and a specific portion of the Lazio region, located south of the lower course of the Tiber River and extending east as far as Praeneste (modern-day Palestrina). As the Roman Empire expanded, Latin became the dominant language of conquered territories and subjugated peoples across the western part of the imperial system. It reached its maximum geographical extent under the Emperor Trajan, specifically following the acquisition of Dacia in $107$ d.C.

In the eastern portion of the Empire, Latin never managed to supersede Greek, which maintained a stronger status as a cultural language. However, Latin remained the language of administration until the era of Justinian, who reigned as the Eastern Emperor from $527$ to $565$ d.C. Notable evidence of this is the grand legislative work of Justinian, the Corpus iuris civilis, which was written in Latin. In the Balkan Peninsula, a clear linguistic boundary known as the 'Jireček Line' (named after the scholar who identified it) separated the northern Latin-speaking area from the southern Greek-speaking area. This line ran from the city of Lezhë (Alessio) on the Adriatic, through Sofia, to the Black Sea along the course of the Danube. In the Mediterranean, coastal Africa west of Carthage was Latinized, while Greek dominated to the east and throughout the Asian provinces.

Written vs. Spoken Latin and the Comparative Method

Direct documentation of Latin is almost exclusively limited to written, primarily literary texts. However, the Romance languages are continuations not of this literary standard, but of the spoken language. All the evolutionary phenomena discussed in linguistic history are inherent to the spoken form. Documentation for spoken Latin is indirect and comes from several sources: anecdotal evidence or remarks by authors about their own speech or the speech of others; observations made by grammarians; and texts that deviate from literary norms for expressive reasons or due to their technical nature (such as treatises on practical subjects).

Linguists also reconstruct spoken forms by comparing correlated forms across various Romance languages. For example, by examining Portuguese ficar ('to stay') and fincar ('to drive in'), Galician ficar, Spanish fincar (obsolete) and hincar ('to thrust'), Catalan ficar, Provençal ficar, French ficher, and Italian ficcare, scholars can deduce the existence of a spoken Latin verb $*FIGICARE$. This verb, derived from $FIGĒRE$ ('to fix' or 'to attach'), is not attested in written literature but is necessary to explain the Romance results. While indirect documentation and reconstruction allow for the classification and dating of most phonetic, morphological, and syntactic phenomena, it remains impossible to describe spoken Latin as a single, unified system.

Unity, Divergence, and the Concept of Submerged Latin

Literary Latin was a remarkably unified language regardless of the author's geographic origin. Although ancient observers sometimes claimed to sense a writer's origin—such as Quintilian ($35/40$ to ca. $96$ d.C.) mentioning a certain padovanità or patavinitas in the language of Titus Livy ($59$ a.C. to $17$ d.C.)—it is unclear what these regional flavors actually entailed. In contrast, the Romance languages that continued Latin are clearly distinct, a divergence fundamentally linked to the collapse of political, administrative, and educational structures following the crisis of the Western Roman Empire.

Research into inscriptions allows for the identification of regional linguistic traits outside the literary norm, especially in later phases, though these do not always align perfectly with modern Romance divisions. St. Jerome, in $386$ d.C., famously noted that "Latin changes every day according to where it is spoken and with time" («cum […] ipsa Latinitas et regionibus cotidie mutetur et tempore»). Scholar Alberto Varvaro proposed the hypothesis of 'submerged Latin' ($latino sommerso$), suggesting a level of language that was censored in writing because it was viewed as a sign of ignorance. This submerged level contained the innovations and regional diversifications that only became visible in documentation from the $9^{th}$ century onward.

Social Stratification and the Terminology of Vulgar Latin

While the Western Empire's cohesion lasted, the cultured linguistic norm of Rome exerted a powerful unifying force. This was supported by centralized power, the concentration of cultural production in Rome, an excellent communication system, an economy centered in the capital, and a widespread, efficient school system. These factors suppressed regional tendencies to innovate. Linguistic diversification was largely confined to lower social levels and informal styles. At higher levels of exchange, Latin remained homogeneous, governed by a coordinated norm.

'Vulgar Latin' (from Latin $VULGUS$, meaning 'the mass' or 'the people') is a conventional label for the language that differed from the cultured norm and contained the innovations seen in Romance languages. However, spoken Latin was not exclusively the language of the lower, uneducated classes; it was a continuous reality that spanned all social strata. Even highly educated people used informal expressions, constructions, and pronunciations in colloquial settings that they would censor in formal or written contexts. Consequently, the term 'Late Spoken Latin' is often more appropriate when discussing the centuries immediately preceding the birth of Romance languages. The term 'Pre-Romance' or 'Proto-Romance' refers to the state of Latin reconstructed through the historical-comparative method looking backwards from Romance languages ($diachronia retrospettiva$). In contrast, 'Vulgar Latin' typically refers to the perspective of looking forward from documented Latin toward its Romance outcomes ($diachronia prospettica$).

Sources for Knowledge of Vulgar Latin

In antiquity, there was no interest in transcribing live speech, so 'spoken Latin' texts in the modern sense do not exist. Even writings by the less educated were influenced by the formal rules of the school system. Significant sources include the personal letters of Cicero, which demonstrate a colloquial but still cultured register. Practical, non-literary language is found in papyri and ostraca (shards), particularly soldiers' letters from Egypt, such as the $14$ letters from the family of the legionary Claudius Tiberianus (early $2^{nd}$ century d.C., with one referencing events in $115$).

Inscriptions collected in the Corpus Inscriptionum Latinarum ($CIL$), which currently spans $17$ volumes, are vital. These include carved formal texts (epitaphs, dedications), painted inscriptions, and graffiti (especially from Pompeii, preserved by the eruption of $79$ d.C.). Errors in these texts provide insight into actual pronunciation. For example, the spelling $HOCTOBER$ instead of $OCTOBER$ is a hypercorrection; the writer did not pronounce the [h] but knew it was sometimes used in writing and mistakenly applied it where it did not belong. Similarly, writing $SCRIPIT$ for $SCRIBIT$ indicates a confusion between [b] and [p]. Other sources include the 'defixionum tabellae' (curse tablets), the archaic and popular language in Plautus (died ca. $185$ a.C.), and the expressive characterization of speech in Petronius's Cena Trimalchionis ($1^{st}$ century d.C.). Technical works on agriculture and veterinary medicine, such as the Mulomedicina Chironis ($4^{th}$ century d.C.), also preserve non-literary vocabulary.

Metalinguistic Sources and the Appendix Probi

Metalinguistic sources consist of observations by authors and grammarians about the language itself. St. Augustine ($354$ to $430$ d.C.) famously noted that it was better to use the non-classical $OSSUN$ ('bone') than the classical $OS$, to avoid confusion with $OS$ ('mouth'), observing that African ears no longer distinguished between long and short vowels. This indicates that the distinctive value of vowel quantity was lost in Africa by the early $5^{th}$ century while still persisting in Rome. Additional information comes from grammarians like Sergius, who noted phonetic qualities (open vs. closed) of vowels in the $4^{th}$ century.

A crucial document is the Appendix Probi, a list of $227$ prescriptions in the form "X non Y" found in a codex from Bobbio (late $7^{th}$ to early $8^{th}$ century). It documents various evolutionary phenomena: syncope ($SPECULUM$ non $SPECLUM$), the use of diminutives and diphthong changes ($AURIS$ non $ORICLA$), the closing of unstressed vowels leading to palatalization ($VINEA$ non $VINIA$), and hypercorrection ($OSTIUM$ non $OSTEUM$). Recent scholars, such as Asperti and Passalacqua ($2014$), date the core of the Appendix to the mid-$5^{th}$ century d.C.

The Role of Substrata in Linguistic Change

Substrate languages are those to which Latin replaced in various regions. This process involved periods of bilingualism and diglossia before the local language was abandoned (with the notable exception of Basque). Some scholars, following Graziadio Isaia Ascoli, attributed Romance phonetic changes to the influence of these substrata. While these theories are often debated, the lexical influence of substrata is certain.

Key substrate influences include:

1. Etruscan and Italic: The 'gorgia toscana' is sometimes attributed to Etruscan. Italic languages like Oscan and Umbrian (documented in the Iguvine Tables) likely influenced the assimilation of $-MB-$ to $-mm-$ and $-ND-$ to $-nn-$ in southern dialects.

2. Greek: Highly influential in Southern Italy (Magna Graecia), Marseille, and Rome. Cicero used Greek models to create Latin philosophical terms like $QUALITAS$ (from Greek $\pi\omicron\iotaό\tau\eta\varsigma$).

3. Celtic (Gaulish): Spoken in Cisalpine and Transalpine Gaul. It left lexical traces and has been credited with the change of tonic $A$ to [e] in French (e.g., MARE > mer), the transition of Latin $-CT-$ to [jt] (e.g., FACTU > fait), and the shift of Latin $U$ to [y].

4. Iberian/Basque: The loss of initial $F-$ in Castilian is often attributed to an Iberian substrate.

5. Balkan Substrata: Illyrian, Thracian, and Dacian underpin the development of Romanian, contributing approximately $80$ words related to local flora, fauna, and geography.

Christianity and the Shift in Linguistic Ideology

Christianity played a fundamental role in linguistic innovation. Following the Edict of Milan ($313$ d.C.) and the Edict of Thessalonica ($380$ d.C.), it became the official religion. Christian authors promoted a new linguistic ideal: $humilitas$ or the sermo humilis ('humble style'). St. Augustine argued that it was better for grammarians to criticize a speaker than for the people not to understand them («melius est reprehendant nos grammatici quam non intelligant populi»).

Despite radical views held by some, like St. Jerome's early vow to stop reading 'secular books,' the Church ultimately decided to preserve the ancient classical tradition as a tool for Christian culture. This decision, championed by St. Augustine, ensured that pagan Latin works continued to be copied during the transition from papyrus rolls to parchment codices. Most surviving ancient texts exist today because of the editorial activity and copying that occurred in ecclesiastical schools and during the Carolingian era ($8^{th}$ to $10^{th}$ centuries).

Empire Fragmentation and the Germanic Superstrata

The unity of Latin disintegrated alongside the Roman Empire. Significant milestones included the transfer of the capital to Constantinople ($330$ d.C.), the final division into East and West ($395$ d.C.), and the sack of Rome by the Visigoths ($410$ d.C.). By $476$ d.C., the Western Empire effectively collapsed with the deposition of Romulus Augustulus. Various Germanic kingdoms emerged: Visigoths in Spain and Southern France, Franks in Northern Gaul, Ostrogoths and later Lombards ($568$ d.C.) in Italy, and Vandals in Africa and Sardinia.

Germanic languages acted as superstrata: languages that overlaid Latin but were eventually abandoned in favor of it. Latin remained dominant due to its written tradition, administrative necessity, and its status as the language of the Church. However, Germanic influence was profound, especially in Northern Gaul where Franconian was spoken by the court of Charlemagne ($742$ to $814$ d.C.). The French language shows the strongest Germanic influence, such as the introduction of 'aspirated h.' Many Germanic words entered the spoken Latin of the Early Middle Ages, covering areas such as warfare, law, and daily life. The collapse of the school system and the ruralization of society meant that regional varieties evolved independently without a central controlling norm.

The Impacts of Arabic and Slavic Superstrata

In $711$ d.C., the Umayyad conquest of the Iberian Peninsula brought Arabic influence to the region. While the Romance-speaking mozarabic population survived in Al-Andalús, the northern Christian kingdoms (Asturias, León, Castile, etc.) began the Reconquista. This process created a 'vertical' linguistic distribution in Iberia: Galician-Portuguese in the west, Castilian in the center, and Catalan in the northeast. Arabic greatly influenced the lexicon, particularly in science, commerce, and agriculture, facilitated by bilingualism and cultural centers like the courts of Alfonso X of Castile and Frederick II in Sicily. Sicily was captured by Arabs in $827$ d.C. and remained under their control until the Norman conquest ($1061$ to $1091$).

In the East, following the Roman withdrawal from Dacia in $271$ d.C., Slavic peoples began settling in the area in the $6^{th}$ century. Unlike the Germanic superstrata in the west, the Slavic superstrata functioned alongside Old Church Slavonic, which became the language of the Church and culture. Consequently, Romanian remained an exclusively spoken language until the early $16^{th}$ century (the first surviving document is from $1521$). Despite its Latin core and grammatical structure, Romanian is heavily characterized by a vast Slavic vocabulary.