You are on page 1of 7

Leah Wild

lw52

Without being able to hear an ancient speaker, we have no idea how a dead language
would have sounded. Discuss with reference to Latin.
As McCullagh states in his introduction to Latin phonology, there is a popular misconception that
given the absence of native speakers, we can have little to no clear idea of their phonology
(McCullagh, 2011). The fact that he calls it a misconception indicates his thoughts on the matter, and
he indeed goes on to detail how we can construe how a spoken language would have sounded from
only written sources. Much of Latin literature is intended to be heard, such as Senecas plays which
were intended to as recitals, the preserved forms of Ciceros speeches in court, and of course poetry,
which can only really be understood when spoken aloud. Therefore, the question of whether or not we
can ever truly recreate how something was intended to be pronounced is one intensely pertinent to our
appreciation of Latin literature. We know some things about how Latin was pronounced: even in
speaking Latin with an Italian accent can give us an albeit limited understanding Latin vernacular. It is
possible to get even closer than this, by deep analysis of the source material, and by utilising other
linguistics methods. We do not therefore, necessarily need an ancient native speaker in order to know
how the language would have sounded. However, there are limits to our understanding. It can be
tricky to pin down when a sound changes, and alterations in pronunciation over time are always
difficult to track, something further complicated by issues such as provincial accents, idiomatic
speech, as well as the fact that the majority of written Latin records the speech habits of a tiny
minority of elite Romans. We shall begin by understanding first how it is we can know anything at all
about Latin. By understanding the methodology of reconstructionist linguists, we will be able to
examine the strengths and limitations of their approaches. Secondly, we shall examine cases where we
have a fairly comprehensive understanding of how Latin would have sounded to a native speaker.
Finally we will move onto cases where our understanding is more limited, and assess how far this
limit prevents us from being able to say how something was pronounced. My aim is to show that
while we can gain a working understanding of Latin speech without needing to listen to an ancient
Roman speak, we are still removed from a fuller understanding which would be gained by having
access to a spoken source.
The methodology of the linguists is what allows us to say that we know anything at all about how
Latin sounds. If it were not for the ability of a linguist to decipher the speech patterns behind the
writing patterns, we would not be able to tell anything at all. Fortunately, languages are primarily
spoken things, and so when things are written down, they are always done with reference towards the
spoken word. As Quintilian notes, Ego nisi quod consuetudo obtinuerit sic scribendum quidque
judico, quomodo sonat (But if I which practise has prevailed so I write what I judge the manner in
which it sounds) [Id. ib. iv. 7.] (Lord, 1894). From the written record, linguists can infer many things
about how ancient vernacular might have functioned. For instance, the existence of spelling mistakes
shows us how Romans thought things ought to be spelt based off how they sounded. Looking at a
corrected scribes tablet, for instance, allows us to see what common mistakes were, and thus what
were common modes of pronouncing things. Two letters that are often mixed up suggest similar
sounds; conversely, two letter which are never confused are likely to have totally different sounds.
Munro argues that if the spelling of a word changed then this indicted that there had also been a
systemic change of pronunciation also, as seen in the fact that /AE/ was never misspelled /E/, but
instead misspelled as /AI/, indicating its pronunciation was closer to the latter than the former
(quoted in: Lord, 1894). However, when we do start seeing /AE/changing to an /E/, we are able to
infer that there has been a shift in the popular pronunciation of the word.
We know exactly how Cicero or Quintilian did or could spell... I have the
conviction that in their best days philological people took vast pains to make the
writing exactly reproduce the sounding; and that if Quintilian or Tacitus spelt a
word differently from Cicero or Livy, he also spoke it so far differently.- Monroe
(Quoted in: Lord, 1894)
1

Leah Wild
lw52

The only issue with this is that writing is inherently conservative, lagging behind the developments of
speech. Given that our understanding of Latin is entirely dependent on this manuscript tradition, this
can prove frustrating, as written convention fails to keep pace with linguistic developments. Even then
we cannot rely on the manuscript to be fully accurate, as often it is preserved in Medieval scholars
texts, and may have therefore been subject to revisionist corruption and policing by Medieval and not
Classical standards.
Linguists also use the writings of ancient grammarians, in order to gain an idea of how people who
spoke Latin thought that it ought to be pronounced. This offers a direct account of how Latin worked,
from people who were themselves native speakers. However, while offering valuable insights, this
also raises some issues. These grammarian texts are often written much later than the Classical Latin
they purport to report, so are limited in their capacity to tell us about the Latin of Ciceros Rome.
Furthermore, they lack the descriptive framework which modern linguists have to work with: there
was no International Phonetic Alphabet, and therefore descriptions of sounds can be somewhat vague,
such as the description of a /R/ sounding like a dog growling (Thompson, 2015). Moreover,
grammarians desire to preserve a particular kind of speech, which is limited to the upper classes and is
in many cases reactionary and conservative, and not reflective of the vernacular of the time, as we can
see in the development of Romance languages. Puns, word play and ancient etymologies are also ripe
grounds for discovering what the Romans through things sounded like. By looking at this Roman
statements of what things sounded similar enough to be noteworthy, we can compare and contrast the
different spellings and come to realisations of pronunciation based off this inferred similarity.
(McCullagh, 2011)
Further evidence for how Latin would have sounded can be found surviving members of its family,
which retains some of its characteristics. Romance languages, like Italian, Portuguese, French and
Spanish, which all descended from Latin, and some of the phonemes of Latin are preserved in them.
The issue with this is deciphering which sounds are left over from Latin and which are later
developments. Indeed there can be a lot of conflict between what is preserved in vulgate Latin and
what is written in the text of the grammarians. This can be partially attributed to the kinds of language
grammarians wished to preserve (the language of the ancient upper-classes) and the kind of language
spoken by ordinary people, Vulgate Latin, which promogulated and passed on into Romantic
languages. We can also use evidence from internal Latin structures, such as meter, to inform our
understanding of how Latin would have been spoken. If something is required to be a certain length in
order to fir the meter, then we can couple this with other rules to see how it must function. For
instance, /H/ hardly registers, metrically speaking, suggesting its role is similar to the breathing in
Greek. Furthermore, transcriptions of Latin into other writing systems, such as Greek, are informative
as to how the Latin would have been pronounced. By comparing the two, we are able to deduce which
of the phonemes match up and which are unique. Finally, there is also ample opportunity for the
linguists to make use of our extensive modern knowledge of the typical phonological patterns and
developments found across languages (McCullagh, 2011), in order to infer the process of
development. Linguists usually combine multiple methods when attempting to discern how something
ought to have sounded. And while each of these methods present their own unique challenges and
unreliabilitys, from this breadth of evidence, certain features of Latin vernacular can be reliable
detrained. Now we know exactly how linguists go about discovering the vox latina, we can explore
it in more in-depth examples to determine the extent of its merit.
One particularly fascinating case for arguing that we know how Latin would have been spoken is by
examining what we know of the pronunciation of /C/, /K/ and /Q/. This is a case that is particularly
convoluted, and yet reaches a rather satisfying phonological conclusions. A phoneme is a perceptually
distinct unit of sound in a specified language that distinguish one word from another. These can be
built up into a linguistic system to transmit meaning. In Latin, almost every phoneme has its own
character to represent it (McCullagh, 2011).
2

Leah Wild
lw52

Linguists are convinced that the /C/ represents the so-called hard c, as in King and not a soft c,
as in century. While it is a much agreed upon fact, with Allen stating that in the grammarians there
is no suggestion that C is anything other than a velar plosive (Allen, 1978), this does not account for
the letter /K/, which also (arguably) represents a hard c. If the two letters represent one and the
same phenome, why bother with having two such letters at all? Allen sites the fact that in early Latin
inscriptions, /C/ seems to have been used before /I/ and /E/, whereas /K/ is used before consonant and
the letters /A/, /Q/, /U/ and /O/ (e.g. Kalendae, kaput, liktor). This indicates variation of pronunciation
in accordance with the environment. However, many ancient grammarians argue that an rursus aliae
(sc. litterae) redundent ut /K/ et /Q/? (Or are other letters redundant, like <K> and <Q>?)
(Thompson, 2015), and argued that K perspicuum est littera quod uacare possit, et Q similis,
namque eadem uis in utraque est. (<K> is clearly a letter we could do without, and <Q> likewise,
since they both have the same value.) (Thompson, 2015). This seems to indicate the fact that there
was indeed very little difference between /C/ and /K/, to the point of interchangeability. We find that
later Latin drops the use of /C/ and /K/ in different words, as Monroe earlier argued it ought, assuming
that this is because it has proven unnecessary and unphonemic, leading to the /C/ being
subsequently generalised in all environments, except in the consonantly combination /QU/ (Allen,
1978)
Additionally the letter /G/ is invented to fulfil the need for a phoneme for what is effectively a
voiced /C/, or guh sound. The Etruscans, from whom the Romans took their alphabet, clearly had
no such voiced k and therefore used the Greek gamma for their voiceless velar plosive, /C/ (Allen,
1978). This is why we sometimes see inscriptions written like VIRCO for virgo, because there
was no written way to express this spoken unit. This is confirmed by Diomedes, who speaks of /G/ as
a new consonant: [Keil. v. I. p. 423.] G nova est consonans, in cujus locum c solebat adponi, sicut
hodieque cum Gaium notamus Caesarem, scribimus C. C., ideoque etiam post b litteram, id est tertio
loco, digesta est, ut apud Graecos posita reperitur in eo loco. (Lord, 1894). The letter /G/ was
allegedly invented by Spurius Carvilius Ruga (3rd cent. BCE), and is simply the letter /C/ with an
additional stroke (Thompson, 2015). The origin of /g/ being written as /C/ is preserved in
abbreviations of names, such as Gaius and Gnaeus being noted as G. and Gn., respectively
(Allen, 1978). In the case of /C/, /K/, /G/ and /Q/, we actually have a pretty good idea of how ancient
Latin speakers would have pronounced this. We have multiple sources who all agree with each other
making the same point, which can be compared to other language, and gives us a lot of accuracy, and
the ability to declare with confidence that we are reasonable sure that we could say how the ancient
Romans would have pronounced these particular letters in given situations. Without ever hearing an
ancient speaker, we can confidently claim to have knowledge of how they would have pronounced
these particular phonemes.
As certain as we can be about the conclusions reached about the pronunciation of /C, K, G/, there are
also cases where pronunciation of the written word is more uncertain. Instances of this will be
explored through the next section. These challenge the notion that we can fully emulate Roman
speech, whatever that has come to mean. The letter /H/ has been the subject of satire, and is often
subpoenaed into words where it never originally belonged, leaving us rather confused as to when
exactly it was supposed to be produced, and in what way this was intended to happen. /H/ begins as a
diagraph used when transliterating Greek , , , into /PH/, TH, KH/ (Thompson, 2015). This
replaced the earlier system of writing Greek aspirated plosives using only a single cognate Latin
plosive letter. Allen states that inscriptional evidence for , , being transliterated as /PH/, TH,
KH/ began in the early second century, and that is was only after this that the aspirated plosive began
to work its way into native Latin words, notably pulcher for what had once been pulcer (Allen, 1978).
Allen also notes these aspirated plosives mostly occur in the vicinity of a liquid consonant like /L/
or /R/, suggesting that this is a linguistic development unique to Latin, rather than a mere passing over
from Greek (Allen, 1978). Cicero lived at this time of tumult when the aspirated /H/ was coming into
fashion. He claimed that he was used to only sounding out the aspirate with vowels, but conceded to
3

Leah Wild
lw52

popular demand to use it as a consonant and with this scientiam mihi reservavi (kept his learning
to himself) [Cic. Or. XLVIII. 160.] (Lord, 1894). Further argument for this aspirated /H/ is found in
that it does not make position (Allen, 1978), which means that it is not counted as part of the
metrical scheme. It also allows for elision, for example in Virgils Eclogue 10.1 extrem(um) hunc,
Arethusa, mihi concede laborem (Allen, 1978). Because of this, /H/ was prone to being dropped in
speech, in the same manner as it is dropped in certain English dialects, such as Cockney. This is
reflected in the written form, where words like nihil are also found written as nil, where the latter
has totally lost its /H/ from the written form as surely as it is lost it in the spoken.
However, the loss of the /H/ also subjected the letter to a form of hypercorrection from which we can
again draw parallels with English. In English, the debate rages over whether the letter /H/ in its
solitary form ought to be pronounced ache or hache. While the former is grammatically accurate,
the latter came about as a form of hypercorrection: because it is symptomatic of lower-class speech
forms to drop ones /H/s, some people feel the need to insert them into places where originally there
had been none , this faux-pas ironically revealing a certain level of ignorance and pretentiousness.
Like Hyacinth Bucket in Keeping Up Appearances, their desperate attempts to correct their
speech actually leads to a kind of hypercorrection, which in some cases catches on. Much in the same
way that Roy Clarke mercilessly mocks Hyacinths attempts be seen as more upper-class than she
really is, Catullus, in his 84th poem, mocks Arrian for his hypercorrection, stating that cum quantum
poterat dixerat hinsidias (he blows his hs most hartificially, Peter Wigham translation)
(Thompson, 2015), after listing words which contain excessively aspirated phonemes, such as
chommoda for commoda and hinsidias for insidias. It is therefore possible to argue that /H/,
while mostly a mark of aspiration, as Victorinus says that Nigidius Figulus so regarded it, [Mar. Vict.
I. iv. 5.] Idem (N. F.) h non esse litteram, sed notam adspirationis tradidit. (Lord, 1894). Allen
bravely offers the guidelines to pronounce Latin /H/ wherever he finds it in his modern texts and
that that will mean that the reader is following perhaps with even greater consistency than the native
speaker, the habits of at least the most literate levels of Roman classical society (Allen, 1978). There
can be no clear consensus, as it seems that even the Romans were conflicted in its proper usage, and
thus while we can know some instances of Roman pronunciation, we run into significant difficulty
when attempting to come up with hard and fast rules about the nature of /H/.
When examining another case for controversy, that the letter /U/ we gain run into a lack of consensus
as to pronunciation. While we manage to maintain a surprising amount of clarity and relatively little
guess work, while we have consensus that the vowel and consonantal forms were markedly different,
we cannot quite work out the manner in which the consonant form of /U/ ought to have been
pronounced. The /U/ consonant and the /U/ vowel Videntur tamen i et u cum in consonantes
transeunt quantum ad potestatem, quod maximum est in elementis, aliae litterae esse praeter supra
dictis; multum enim interest utrum vocales sint an consonants, as Priscan argues [Keil. v. II. p. 13.]
(Lord, 1894). The /U/ vowel is generally thought of as a semi-vowel, like the English /W/. Evidence
for the /U/ as sounding like w is seen in borrowings from Latin, which have turned the /U/ into a
/W/ example Gothic "wins" meaning 'wine' was borrowed from Latin "vinum" (Covington, 2010).
Furthermore, Nigidius Figulus, when giving a description of pronouns, argues that ones where the lips
protrude refer to others, as if pointing with ones lips, and pronouns that refer to oneself do not point
outwards at all (Allen, 1978). Despite the dubious nature of the claims, this does indicate that vos, a
pronoun used to refer to others, cannot be pronounced with a vee sound, because this does not cause
lip protrusion, strengthening the argument for a double u sound for the /U/ consonant, because of
the lip rounding required to form the sound. This bilabial, semivocalic articulation (Allen, 1978)
does not mean that there is not a link between the two sounds of the vowel and consonant forms of
/U/. The two are, to some extent, interchangeable, as proven in poetry. Allen cites the example of the
disyllabic silua which can be trisyllabic, and the trisyllabic genua, which can be disyllabic. This
comes about by a switch from the vowel form of /U/ to the consonant, and vice versa, from their
conventional [sil.wa] and [ge.nu.a], to [si.lu.a] and [gen.wa] (Thompson, 2015). Horace makes use of
4

Leah Wild
lw52

this in Epode VIII. 2 where he writes silva with the vowel form of /U/: Nivesque deducunt Jovem,
nunc mare nunc silu
(Lord, 1894). This interchange of function is used as metrical evidence for the
differing pronunciation. Further evidence for pronunciation can be found in Greek transliteration,
which writes the consonantal /U/ as /ou/, such as in Greek transliterations of the name of the Emperor
Vespasian. In Latin it is written as /VESPASIANUS/ whereas in Greek transliterations this is spelt
Ouespasianos. This implies that the initial /V/ was pronounced in a way that is closer to
Wespasian than 'Vespasian (Harris, n.d.). Moreover, Ciceros famous pun in de Divinatione
offers evidence for the similarity of sounds. Cicero tells of when Marcus Crassus, who was setting off
on an expedition which would go poorly, was said to have heard someone crying Cauneas! (i.e.
Figs from Kaunos!) (Allen, 1978). Cicero however suggests that he ought to have heeded the
warning in these words, and instead heard them as Cave ne eas! (Beware! Do not go!). If the two
forms are so similar as to be viable material for a pun, then their pronunciation is clearly in some
ways similar, as the consonantal /U/ of cave (presumed consonantal because it comes between two
vowels), and the vowel of /U/ in Cauneas, can be plausibly conflated.
As strong as this evidence seems to be, the association of the /U/ consonant with the early Greek
digamma casts some doubt on the conclusions drawn. The digamma symbol stood for a lot of things
in Greek, and in Latin had been altered, and so now stood for the fricative eff sound. However,
when the Romans transliterated the digamma, they represented it as /VAU/ (Lord, 1894). We hear
from Marius Victorinus that [Keil. v. VI. p. 23.] Fautem apud Aeolis dumtaxat idem valere quod
apud nos vau cum pro consonante scribitur, vocarique et digamma (Lord, 1894). Which links
the to the digamma. sounds like the English v, but Lord claims that the office of the Greek
digamma was apparently manifold (Lord, 1894), making it hard to pin down which of its apparently
many sounds the Romans associated with the consonantal /U/. We may therefore decide to abandon
this particular avenue of conceptualising /U/. However, it was something that seemed to preoccupy
the ancient grammarians. Quintilian says that even if the form of the Aeolic digamma is rejected by
the Romans, yet its force pursues them: [Quint. XII. x. 29.] Aeolicae quoque litterae qua servum
cervumque dicimus, etiamsi forma a nobis repudiata est, vis tamen nos ipsa persequitur (Lord,
1894). While most modern linguists believe that the /U/, when a consonant, was pronounced like a
w, there are arguments against this, such as the fact that the w sound is not found in later Greek,
nor is it present in the Romance languages that developed from Latin. Lord points out that In French
we find the Latin u consonant passing into f, as ovum into uf; novem into neuf (Lord, 1894). Lord
also points out that The w sound is not only unfamiliar but nearly, if not quite, impossible, to the lips
of any European people except the English and uses this to conclude that the labial pronunciation of
the consonantal /U/ is not necessarily the pronunciation of the ancient speakers (Lord, 1894). While
there is therefore some disagreement over the pronunciation of the /U/ consonant, linguists have
reached a general consensus, which is supported by the majority of the sources. We can merely offer a
working opinion of what this would have sounded like from a native Latin speaker.
It is therefore possible to conclude that while we can have a fairly comprehensive picture of the Latin
phonetic landscape, there are some precise details that still elude us. Therefore, we cannot as of yet
state that we fully understand how Latin would sound when spoken by a native. Furthermore, our
search, we draw our understanding from a wide breadth of sources: we use inspiration evidence,
graffiti, treatises of grammarians, anecdotal word-play, developments from Indo-European and later
developments into Romance languages. However, these things all record different dialectical uses of
Latin, some being the preserve of the upper classes, especially the written format, but things such as
the transition of Vulgate Latin to Romance Latin are necessarily far more to do with Latin as spoken
by the common Roman, as indeed its description as vulgate suggests. There is an admitted bias
towards striving to understand the limited vernacular of educated speakers of Latin during the
Classical period (McCullagh, 2011). Attempting to reconstruct the Classical Latin that we want to
from these sources, while workable to a certain extent, is necessarily lacking. Indeed, the entire
premise of the argument, that we can discover what an ancient speaker would have sounded like is
5

Leah Wild
lw52

inherently based on the notion that there is going to be a correct kind of ancient speech, trapped in a
very specific, probably Ciceronian time-frame. However, much can be elucidated from the sources
that do survive, by means of forensic interrogation of the existing evidence. This is enough to provide
us with a working model for standard Latin pronunciation. Gaps can be filled in phonological
extrapolation, until we come the fullest understanding. To the greatest extent we can vocalise Latin to
a respectable degree, and the major arguments tend to fall down on points of minutiae. We can glean a
great deal form phonological data, and thus to conclude with the words of Lord, what cannot be fully
known or perfectly acquired does still not prevent our perceiving (Lord, 1894).

Leah Wild
lw52

Works Cited
Allen, W., 1978. Vox Latina. Cambridge : s.n.
Covington, M., 2010. Latin Pronunciation Demystified. s.l.:Program in Linguistics, University of
Georgia.
Harris, W., n.d. THE PRONUNCIATION OF LATIN: Ancient and Modern Pronunciations. [Online]
Available at: http://community.middlebury.edu/~harris/LatinBackground/Pronunciation.html
[Accessed 28 02 2016].
Lord, F., 1894. The Roman Pronunciation of Latin: Why We Use It, and How To Use It. Gutenberg
text ed. Boston: Ginn and Co..
McCullagh, M., 2011. The sounds of Latin: Phonology. In: J. Clackson, ed. A Companion to the Latin
Language. Oxford : s.n., pp. 83-91.
Thompson, R., 2015. The Sounds of Greek and Latin: Reconstructing a Dead Language (lecture
handout). s.l.:s.n.

You might also like