General Principles:

The reader will doubtless be aware that there are several styles of Latin pronunciation which, while in broad enough agreement to make most of them mutually recognisable, differ in many details. This is certainly to be expected in the case of a litterary language which has not had a living colloquial dialect for some fifteen centuries. These styles fall into three major groups:

The worst of all, which warrants mention only because of the annoying frequency wherewith it is to be found, is the shamefully ignorant pronunciation of Latin as if it were modern English. This farcical pronunciation totally negates the utility of Latin as an international language, as it is rarely understood in any real way by those who 'use' it, and hardly understood at all by those whose knowledge of Latin may be quite good, but whose knowledge of English is limited. This is because the vowels of modern English, particularly the so called 'long' vowels, were hopelessly displaced and distorted by a lamentable linguistic disaster known as 'The Great Vowel Shift', which had its beginnings in early Tudor times, but swept through most of England during the reign of Elizabeth I. During the eighteenth and nineteenth centuries this type of pronunciation was indeed very widespread throughout the English speaking world, even among those who could actually read Latin well enough to translate it, but during the twentieth century it became increasingly discredited, and is now hardly more than an embarrassing curiosity of the post-Reformation era.

The so called 'Classical' pronunciation is a much more scholarly and respectable contender, and is fairly readily understood by almost everyone who has learnt to read Latin. At its most extreme, however, it enshrines certain inherent anachronisms by favouring a phonetic system from the pre-classical, or 'Old Latin' period for the reading of classical texts, in effect ignoring important changes in both pronunciation and spelling which occurred at the very start of the 'Classical' period, some two centuries before Christ. Some varieties of 'Classical' pronunciation, because of their concentration on Classical Latin poetry, are inclined to overplay the distinction between long and short vowels, and in doing so are rather shameless about accommodating themselves to the phonetic habits of the native languages of their proponents and students. Much of what is taken today as 'Classical' pronunciation is credited to German scholars of the early nineteenth century.

The Barefoot Abbess favours a system of pronunciation which falls well within what is recognised as the 'Ecclesiastical' school. The reader should be aware, however, that here too, there are a number of minor variations. Some within this camp insist on basing their pronunciation of Latin rather strictly on the pronunciation of modern Italian, while others would try to reconstruct the pronunciation of the Latin Fathers, particularly St. Jerome, who compiled the Vulgate Bible, and others yet are willing to selectively borrow certain features associated with 'Classical' pronunciation in order to find a better match between the written word and the spoken word. The Barefoot Abbess has attempted here to define a system of pronunciation which preserves and balances all of the following principles:

First: That as far as possible a one to one correspondence be maintained between the written characters and the phonemes they represent. Pronunciations which obscure the distinction between written characters or character combinations are thus to be avoided, and pronunciations which clarify such distinctions adopted, even where they might otherwise be a bit anachronistic. Latin is, after all, essentially a _written_ language, and this principle is of great importance in helping students of any written language to learn it in the most accurate and efficient way.

Second: That all Barefoot Abbess pronunciations take as full an account as possible of the _entire_ history of the Latin Language and its development, but in so doing still pay special attention to the Latin of educated Romans at the time of St. Jerome.

Third: That all Barefoot Abbess pronunciations be well within the range of what is recognised as good Ecclesiastical usage.

Phonemes, Allophones, and Alphabets:

Out of the rather vast continuum of possible human speech sounds, the language of a civilised people typically recognises about three dozen phonemes, employing about two dozen in pronouncing its native word stock, and perhaps another dozen in pronouncing words borrow'd from respected and influential neighbours. Each phoneme, particularly the 'native' phonemes, will typically include two or three variants, known as 'allophones', within its recognised range. The allophone one actually hears for a given phoneme depends in part on the dialect, or local 'accent' of the speaker, and in part on the other phonemes which precede or follow it in a given word. It is very important to understand that each language recognises a set of phonemes and allophones unique unto itself, and that two sounds which might be distinguished as distinct phonemes in one language are often heard only as allophones of a single phoneme in another language. When an alphabet has been either created or thoroughly adapted for a particular language, one can expect something very close to an exact correspondence between its characters and the phonemes of that language.

The Latin Alphabet:

The earliest true alphabet of the Mediterranean world was essentially the same 22 letter Semitic alphabet used to write Hebrew up to the time of Esdras. It was also the alphabet of the Phonetians, a people whose language, like that of the Hebrews, was of Semitic origin, but whose influence, as seafaring merchants, was far more widespread in the ancient world, and indeed stretched all across the Mediterranean. The early Greeks learnt the value of writing through trade with the Phonetians, and were the first to adapt the old Semitic letters to the somewhat different needs of an Aryan language.

The Romans essentially borrowed and adapted a western form of the Greek Alphabet, but they borrow'd it not directly from the Greeks, whose language ultimately had the same Aryan origin as their own, but from the Etruscans, a mysterious people with a language that was neither Aryan nor Semitic, and had a phonetic structure that was totally foreign to either family. This linguistic incompatibility made at first for a very clumsy borrowing that made Old Latin difficult to write and difficult to read.

About two centuries before the birth of Christ, however, the Latin language went through several very significant changes, which finally forced the Romans to make some much needed changes in the alphabet they had borrowed, and in the way in which they used it. The Etruscans, who had once dominated Rome, were by then an all but forgotten people, and the Romans were now borrowing heavily from Greek culture, and from the Greek language. The Latin writers of the early Classical period needed a writing system which could accurately represent not only their own language, including its recent changes, but their Greek borrowings as well, and they did a very good job of adapting their alphabet to this purpose. The Barefoot Abbess tries, in all of her recommended pronunciations, to be ever respectfull of the genius of those early Classical scholars, whose work indeed set the linguistic foundations for the Christian West.

Pure Vowels and Diphthongs:

For speakers of most languages the distinction between a pure vowel and a diphthong is simple and obvious, but for those whose first language is English it is anything but. Most of the so called 'long' vowels of English have in fact been expanded into diphthongs, and several vowel combinations originally pronounced as diphthongs have been contracted into pure and simple vowels. If you can sing a series of repetitions of a vowel sound without any change in the position of your tongue, you are singing a pure vowel. If you feel your tongue repeatedly changing position in your moth and hear an alternation of two different vowel sounds, you are producing not a pure vowel, but a diphthong. Diphthongs do indeed occur in Latin, but it is very important that sounds which are actually diphthongs never be substituted for single vowels which are intended to be pure and simple.

Long and Short Vowels: The Changing Musick of the Latin Language:

Of all the foolish leftovers from the infamous 'English' system of Latin pronunciation, the most pernicious and persistent is a misconception of the distinction between long and short vowels. The worst effect of the linguistic disaster which befell the English Language at the time of the reformation was that English speakers became accustomed to talking about 'long' vowels which no longer sounded anything at all like their 'short' counterparts. This is not the case in Latin. Latin vowels are pronounced in exactly the same way whether they are long or short. The long vowels have exactly the same sounds as their short counterparts. The _only_ difference is their duration in musical time. A long vowel, in Classical Latin poetry, is sustained for about twice the duration of a short vowel, but it does not differ from the corresponding short vowel in any other way. Old Latin inherited, and to a large extent preserved, the musical pitch accent of its Aryan parent language, and this system of assigning a higher pitch to some syllables and a lower pitch to others combined quite naturally with the mix of long and short vowel sounds to produce a very musical language. There is evidence that there was also a degree of stress accent, as in later Latin, and in all historical stages of English, but that this always fell on the first syllable of each word and was of secondary importance. At the start of the Classical period, however, there appears to have been a major change in that both the pitch accent and the initial stress accent disappeared, and a new and stronger stress accent settled upon either the penultimate or antepenultimate syllable of each word; depending on whether the penultimate syllable carried the weight of either a long vowel or a short vowel followed by two or more consonants, in which case it would carry the accent; or whether it had only a short vowel followed by a single consonant, in which case the accent would fall on the next earlier syllable, known as the 'antepenult'. It might be argued by some that the replacement of musical pitch accent by stress accent makes a language sound more 'military', more 'masculine', and perhaps more 'adult'. Be that as it may, it is ultimately a mischief. It results in the shortening of any long vowels in syllables earlier than the accented syllable, which is relatively harmless; but it also works to substitute articulatory tension for actual duration as the primary distinction between 'long' and 'short' vowels, and 'long' vowels that have come to be distinguished more by a greater overall tension in the mouth than by a longer duration than their shorter counterparts are always in danger of closing up when a generation comes along that is for some reason particularly lazy in learning its speech habits. This, of course, is exactly what happened to XVIth century English. There is, finally, some question as to whether the distinction between long and short vowels should even be observed at all in Ecclesiastical Latin. Important as it was in Classical Latin poetry, the distinction had never been important enough in prose or in ordinary speech to inspire any consistent attempt to represent it in writing (although practices such as the doubling of long vowels seen in some Old Latin inscriptions, and the use of a mark call'd an 'apex', which resembled an acute accent, seen in many Classical Latin inscriptions, were nonetheless common enough to inform later scholars as to which vovels were of long duration in the pronunciation of those respective periods.) The distinction between long and short vowels, however, had only rarely been of importance in distinguishing one word from another, and had largely disappeared from ordinary speech by the start of the Christian period. Despite the fact that the distinction is more often than not totally ignored in church usage, the Barefoot Abbess would nonetheless encourage any student of Latin to try to learn and observe it, so long as it be observed properly as a matter of vowel _duration_ only. Gaining a sense of what is going on in Classical Latin poetry, is, after all, an extra bonus for the student which comes at little extra cost in learning time.

Open and Close Vowels:

The terms 'open' and 'close' refer to how close the tongue is to the roof of the mouth in the articulation of a given vowel sound. The sounds represented by 'a' and 'æ' are the most 'open' vowels in Latin, as the mouth is most open when they are produced. The vowels 'u', 'i', and 'y' are said to be 'close' in that the tongue comes closest to the roof of the moth when they are pronounced. The vowels 'o', 'e', and 'œ' are of medium closeness.

Front and Back Vowels:

The terms 'front' and 'back' refer to which part of the roof of the mouth the tongue is most closely approaching in the articulation of a given vowel sound. The vowels 'i', 'y', 'e', 'œ', and 'æ' are all front vowels, while 'u', 'o', and 'a' are all back vowels in Latin.

Rounded and Unrounded Vowels:

When producing the vowels 'o', and 'œ' the lips are always 'rounded' into a relatively small circular opening, and they are even more tightly 'rounded' into an even smaller circular opening when producing the vowels 'u', and 'y'. In producing the vowels 'a', 'æ', 'e', and 'i', however, the lips are never rounded at all.


The close vowels 'i' and 'u' can be sounded not only with the long and short durations typical of the other vowels, but can also be sounded with an even shorter duration more typical of consonants. When so sounded they are also pronounced with the tongue even closer to the roof of the mouth, so that a more consonant-like friction is produced. In such cases they are known as 'semivowels'. The semivowel allophones of 'i' and 'u' are typically heard when they are found before other vowels at the beginning of a word, between vowels in the middle of a word, or when they are to be sounded as the final elements of a diphthong.

Closest English Equivalents
to the Vowels, Semivowels, and Diphthongs of Latin:

Simple Vowels:

'A' may be of either long or short duration, but in either case should always sound like 'a' in 'father'. It is sometimes permissible to relax a short, final 'a' to a neutral vowel, like the final 'a' heard in Latin feminine names when pronounced in English; but such a lax, virtually neutral 'a' should never be heard in any other position, and simply pronouncing every Latin 'a' like 'a' in 'father' is arguably better.

'E' may be of either long or short duration, but in either case should always sound like 'e' in 'met'.

'I' may be of either long or short duration as a full vowel, or of even shorter, consonantal duration as a semivowel. When long it should sound much like 'i' in 'machine'. When short it should sound much like 'i' in 'mit'. As a semivowel it should sound like 'i' in 'onion', a sound more typically represented by 'y' in English. Do not be confused by the fact that as a semivowel it is often written as 'j'. The only time it is ever to be sounded like 'j' is sounded in English is in the combination 'dj', which when occurring before a vowel sounds like 'dj' in 'adjutant'.

'O' may be of either long or short duration, but in either case should always sound like 'o' in 'roll'. Avoid stretching a long 'o' into a diphthong like 'ow' in 'snow' or 'ew' in 'sew' and be even more carefull to avoid ever opening a short 'o' into the any of the sounds heard in English words like 'dog' or 'fox' which in Latin would be heard as allophones of 'a' rather than 'o'. Latin 'o' is always relatively close and is always distinctly rounded.

'U' and 'V' are variant forms of the same letter, which may be of either long or short duration as a full vowel, or of even shorter, consonantal duration as a semivowel. It is to be regarded as a semivowel whenever it stands before another vowel at the beginning of a word, when it occurs between two other vowels, when it occurs after 'g','q', or 's' and before another vowel, and when it occurs as the final element in the diphthongs 'au' and 'eu'. When long it should sound much like 'u' in 'rule'. When short it can have a more relaxed sound like 'u' in 'put', but must never be allowed to become so lax as the virtually neutral vowel in 'putt'. As a semivowel it should sound like 'u' in 'quit' or 'suave', a sound more typically represented by 'w' in English. Do not be confused by the fact that 'u' is often written as 'v', and indeed may often be heard pronounced like the consonant 'v' in 'very' in some positions, especially when it stands before a vowel at the beginning of a word. Barefoot Abbess prefers to avoid using this last, and historically latest sound for 'v', in part because it is might also be heard in Latin as an allophone of 'f'; but good Latin usage leaves room for some personal variation here. Do remember, though, that from a Latin point of view there is only one phoneme here, regardless of how many allophones you may use to realise that phoneme in your actual pronunciation.

'Æ' is to be regarded as a single, independent letter representing a single pure and simple vowel. The shape of the letter is intended to indicate that its sound is intermediate between that of 'a' and that of 'e'. Regarded as either a fronted 'a' or an open 'e', it should always sound like 'æ' in 'æroplane', and always be of long duration. It is always long for the historical reason that it resulted from the simplification of an Old Latin diphthong which, until about two centuries before Christ, was spelt 'AI' and pronounced like 'ai' in 'samurai'. Some advocates of 'Classical' pronunciation insist on pronouncing it in this preclassical way despite the change of spelling, citing perhaps the fact that Classical Latin writers used 'Æ' to translitterate the similar diphthong which continued to be written as 'AI' in Greek. The true explanation for this, however, is that this ancient diphthong had undergone almost exactly the same change in most Greek dialects, but that Greek spelling had solidify'd at an earlier time, and was thus more reluctant to reflect such changes.

'Œ' is seen far less frequently than 'Æ', but it has a very similar history, and like 'Æ' is to be regarded as a single, independent letter representing a single pure and simple vowel. This shape of the letter is intended to indicate that it's sound is intermediate between that of 'o' and that of 'e'. Regarded as either a fronted 'o' or a rounded 'e', it is to be pronounced with the tongue in position for 'e' but with the lips rounded as for 'o'. It is always long for the historical reason that it resulted from the simplification of an Old Latin diphthong which, until about two centuries before Christ was spelt 'OI' and pronounced like 'oi' in 'oil'. Like its sister 'Æ', it is also used to transliterate a similar ancient diphthong which continued to be written 'OI' in Greek, even though its sound had changed in the same way as the sound of its Latin counterpart.

It must be noted that although 'E','Æ',and 'Œ' would have still had the distinct values cited above in the Latin of the early Christian Fathers, and (as we can see from the fact that the Anglo Saxons learnt the Latin alphabet with these values when they accepted Christianity) for some time later; by the late middle ages they had become so frequently confused as to be virtually interchangeable. Despite a more recent trend to restore their proper distinction in spelling, they are much less often restored to their proper values in pronunciation. The Barefoot Abbess, however, strongly recommends that students learn to distinguish them in recitation as well was in writing.

'Y' is similar to 'Æ',and 'Œ' in that it is best regarded as either a fronted 'u' or a rounded 'i', pronounced with the tongue in position for 'i' while the lips are tightly rounded as for 'u'. Unlike the other fronted vowels, however, it is not a simplification of an Old Latin diphthong, but a direct borrowing of the Greek letter 'Υ' (Ypsilon) which had this same sound. For this reason it could be of either long or short duration, depending on its duration in the borrowed Greek words wherein it occurred.


A diphthong (or 'glide') occurs when the tongue moves smoothly to transform one vowel into another vowel or semivowel within the same syllable. Although it is typical in the pronunciation of most diphthongs for a whole series of vowel sounds to be produced in unbroken succession, Latin, like most written languages, represents only the initial and final elements. The final element of a true Latin diphthong has always been a close vowel (either 'i' or 'u') taken to terminate in its closest value as a semivowel. Old Latin had a half dozen diphthongs, but at the start of the classical period 'AI', and 'OI', became 'Æ' and 'Œ' as described above, 'OV' or 'ou', originally pronounced like 'ow' in 'snow', became simple 'V' or 'u', and 'EI' originally pronounced like 'ei' in 'neighbour', became simple 'I'. All Latin diphthongs have always been of long duration, and thus the simple vowels that came to replace those that were simplify'd always remained as long vowels. Some of these lost diphthongs, particularly 'EI', may seem to survive into Classical and Christian Latin, but most of these apparent survivors are likely to have never really been diphthongs at all, but to have always been pronounced as two separate vowels in two separate syllables. When 'EI' is found in Christian Latin, the two vowels should, in any case, always be pronounced separately. Only two diphthongs actually survived into Classical and Christian Latin. The most important is written 'AV' or 'au' and pronounced like 'au' in 'landau' with a sound that begins as 'a' in 'father', glides up through the 'o' and 'u' ranges and ends as the semivowel usually represented by 'w' in English, all within a single syllable. The other surviving diphthong, 'EV' or 'eu' is seen much less often, and apart from Greek loan words is actually quite rare. It is pronounced like 'ew' in 'sew', but with the initial 'e' element not only clearly heard, as it is heard in the Standard English of Southern England, but by English standards quite exaggerated. This diphthong thus begins as 'e' in 'met' and glides smoothly through the 'o' and 'u' ranges to likewise end with the semivowel usually represented as 'w' in English, all within a single syllable.

The table below shews the typical transliterations of Greek vowels and diphthongs into Latin, along with the markings used in Barefoot Abbess Gothic to help indicate which Greek vowels underlie the Latin. The Greek vowels have been shewn in a script intended as a compromise between traditional lower case Byzantine and Barefoot Abbess Gothic. Greek vowel equivalents

Diæresis Ë:

A double dot over a vowel, known as a 'diæresis' is sometimes used to indicate that what would normally be read as a diphthong, or even a ligature expressing a single simple vowel, like 'æ', is instead to be read as two separate vowels with the second vowel being of long duration and carrying the stress accent. In Church Latin the most frequent case of this is in names ending in '-ël', the Hebrew particle which means 'God' or 'god', and is always incorporated in the proper name of an archangel. It is also seen in some other proper names like 'Israël'. This final particle always takes the accent.


Of all the classes of consonants, the liquids, 'L' and 'R', are the most similar to vowels, and they behave very much like semivowels. They present problems, however, in that they subsume much larger numbers of allophones than do other phonemes, and thus vary greatly from one language to another. Latin, like most languages allows liquids to begin or end words and to follow almost any consonant when a vowel then follows, but unlike some languages does not allow a liquid to carry a syllable as if it were a vowel. The liquids have a curious metrical property in that when a liquid follows another consonant the resulting pair can be counted as either one or two consonants in determining the metrical weight of a syllable and thus, in some cases, allow the stress accent to fall freely on either of two syllables according to the antepenultimate law. Except as noted below in the case of 'rh', liquids are voiced when initial. They are also voiced when following a vowel or a voiced consonant, but become unvoiced after unvoiced consonants.

'L' should be noticeably 'clear' in all positions, which means that of the two 'l's in 'little', the first is a better model for Latin 'l' than is the second.

'R' should be distinctly heard as a tongue point trill in all positions. It is most definitely never silent, and it never alters the quality of the vowels before or after it. It should thus be noted in particular that 'er' is always to be sounded as 'er' 'merry'. Latin uses the combination 'RH' to represent the aspirated liquid heard in Greek words. It is best to learn this by pronouncing it as if it were 'hr', and thus to leave it unvoiced.


The nasals, 'M' and 'N', are in many ways similar to the liquids, and thus also similar to semivowels in their behavior. Although the nasals do not share in the special metrical freedom granted to the liquids in combination with preceding consonants, they are allowed in Latin to form such combinations with essentially equal freedom. Nasals, however make their most intimate combinations with following consonants, and these combinations result in 'M' and 'N' sharing a whole family of nasal allophones between them, and indeed in such situations loosing much of their distinction as separate phonemes.

'M' is pronounced as a bilabial nasal, exactly as in English, whenever it is followed by any vowel, or by 'b', 'l', another 'm', 'n', 'p', 'ph', or 'r'. 'M' is fully compatible with these phonemes, but the use of various prefixes and suffixes in Latin often causes the phoneme which is normally spelt, and realised as 'm' to be followed by a consonant wherewith it is not compatible. In such cases it is usually, but not always, respelt as 'n', but even when such an 'm' remains in spelling, it is usually pronounced as one of the allophones associated with 'n'. Final 'm' began to weaken in pronunciation during the Classical period. By early Christian times it had generally given up its consonantal value in favour of a nasalisation of the preceding vowel, and as spoken Latin degenerated further into ignorant local dialects, it was typically lost entirely. Educated people who continued to use the written form of the language, however, reacted against this laxity by restoring final 'm' to its full consonantal value, and virtually all schools of 'Ecclesiastical' Latin pronunciation are carefull to follow in this practice, which for English speakers is most natural in any case.

'N' is pronounced as a dental nasal, with the tongue actually touching the upper teeth, whenever it is followed by any vowel or by 'd', or 't'. This tongue position is a bit forward of the usual position for an 'n' before a vowel in English, but it is much like the position for an 'n' before 'th', like the second 'n' in 'nineth'. The allophone of 'n' most typically heard in English is heard in Latin only when 'n' is followed by 's' or 'z'. When 'n' is followed by the semivowel 'i'/'j', or by the palletised allophones of 'c', 'g', or 't', it is itself palletised, like 'n' in 'onion'. When followed by the nonpalletised allophones of 'c', or 'g', or by 'k', 'q', or 'x', it is pronounced like 'n' in 'ink'.

Other Continuants:

'F', 'H', and 'S' form a sort of a sisterhood in Latin, in that these three continuants are similar to each other, and differ from the other continuants, in that they are most typically voiceless, and always voiceless when they begin words. Centuries before they had any contact with any form of writing, a process began whereby an ancient family of at least four phonemes, which in Greek sorted out into 'Θ' ('Theta'), 'Φ' ('Phi'), and 'Χ' ('Chi'), were in Latin sorted into only two phonemes, for which the characters 'F' and 'H' were used when the Romans adopted the Etruscan Alphabet. (The Etruscan 'F' was taken from the old Greek letter 'F'('Digamma'), which like it's Phonetian ancestor 'Waw', had been used to represent a sound like English 'w' in 'work', but which dissappear'd from most Greek dialects in preclassical times, so that ϝεργων ('work') was reduced to ἐργων. It is unclear exactly what sound 'F' represented for the Etruscans, but they apparently still found it necessary to add another character, which resembled '8' to the end of their alphabet to represent some other type of 'f' sound. This '8' character was also briefly borrowe'd into Latin, and seems to have often been interchangeable with 'F','H', and an 'FH' combination. Only 'F','H' lasted into Classical period, but it is likely that these two phonemes represented a number of allophones, and, for a while at least, shared one or two of them rather indifferently. The extensive borrowing of Greek words into Latin makes it appropriate to note as well that the 'rough breathing' sound which began many Greek words, and which the Romans heard as an 'h' sound, originated in prehistoric Greek as an allophone of the 's' phoneme, and this relationship, although perhaps not fully understand when the Greeks set about to adapt the Phonetian alphabet to their own use, had a profound effect on the development of the Greek alphabet, and thus on the eventual development of the Latin writing system.

'F' in most positions is to be sounded as an unvoiced labiodental continuant much like 'f' in 'fort'. After 'b', 'm' or 'p' it might be sounded as a bilabial, like 'f' in 'comfort', or 'ph' in 'amphibian'. This bilabial allophone may indeed have been the primary allophone of 'F' in preclassical Latin, but Classical Latin does not favour it, and it is best for us to avoid it to avoid confusion with 'PH'. A voiced allophone, like 'f' in 'of', might also have been heard between vowels in some times and places, but is really best avoided to avoid confusion with 'V'.

'H', at the beginning of a word, is an unvoiced glottal continuant, an aspirate equivalent to the 'rough breathing' of Classical Greek, and pronounced like 'h' in 'have'. It would have originated, in Old Latin, as a velar continuant like 'ch' in 'loch', and many still give it this stronger, more forward sound when it occurs between vowels. When the following vowel is an 'i', 'h' is fronted, or palletised, and thus sounded much like 'h' in 'hue' when strongly pronounced. It is possible that in some borrowed words it was used to indicate a glottal stop separating two vowels, but this pronunciation does not survive. 'H' fell into total silence in most of Latin's daughter languages, and in the most extreme 'Italian' styles of Ecclesiastical pronunciation it is left silent; but the Barefoot Abbess nonetheless prefers that it be distinctly sounded, so as to both help the student in the learning of Latin, and the listener in the understanding thereof.

'S' is typically an unvoiced postdental sibilant, sounding exactly like 's' in 'simple'. It has other allophones, but all of its allophones are sibilants. Sibilant are hissing sounds which differ from the sounds mentioned above as allophones of 'h' in that sibilants are produced with the tongue slightly curled to allow the breath to escape through a narrow passage along the middle, while aspirants are produced without this curling. It might be said that 'l' sounds, known to linguists as 'laterals', are the opposite of sibilants in that the tongue is curled in the opposite way when they are produced, so as to allow the breath to escape along the sides of the mouth rather than along the middle. 'S' has a voiced allophone, like 's' in 'rose', but this voiced allophone, which is only heard when a single 's' occurs between two vowels, disappeared from native Latin words during the Old Latin period when a sound change caused it to be reassigned to the 'r' phoneme, and is thus now to be found only in words acquired later from other languages. Note, however, that this situation can appear to occur when an 'open' (i.e. vowel terminal) prefix like 'præ-' or 'pro-' attaches to a word beginning with 's', but the single 's' in such cases remains unvoiced as in the root word. A double 'ss' between vowels is always unvoiced, and indeed also follows the rule that double consonants in Latin are always pronounced as double. Remember that while English almost always prefers that a final 's' be voiced unless it be doubled or follow a unvoiced consonant such as 'p', 't', or 'k', Latin never allows a final 's' to be voiced, and in fact insists on pronouncing a final '-bs' as if it were '-ps' in order to avoid ending a word in a voiced 's'.

The stops and their variants:

The quiescent stops, also known as 'surds' form the most basic category of consonants in virtually all languages. These consonants actually halt the breath stream for a moment of totally dumb silence. It seems a paradox, then, that they are also, in most languages, the largest and most closely distinguished category of speech sounds. How indeed can these essentially silent sounds be heard and distinguished? The answer is that the human ear is by nature acutely attuned to the sound changes that occur during the moments before and after this linguistic shut-down, during the time, that is, when the organs of speech are moving towards and then away from the full stop position. A stop, in fact, has about the same duration as a continuant or semivowel, and although the surd itself be silent, the adsurd and absurd parts of its articulation are very clearly heard. Stops can be articulated with quite a number of different variations. The lips, tongue, teeth, and palate can first of all close upon each other to stop the breath at several different positions. The vocal cords can then either continue to vibrate throughout the articulation to produce a voiced stop -- and it should be noted that this vibration can be heard through the larynx even while the breath is only filling the upper throat without being allowed to escape the mouth -- or they can fall quiet as the mouth parts close into the stop position and stay quiet for another moment as the mouth opens again. This can effectively double the number of stop phonemes. Stops can be modify'd in other ways. In English, for instance, the voiceless stops are typically accentuated by an explosive puff of breath when they occur alone at the beginning of a word or between vowels. This creates a positional allophone whereof English speakers are typically unaware; but these simply plosive stops can sometimes be confused with the aspirated stops which were used as separate phonemes in Classical Greek, and which the Romans heard, and transcribed in borrowed words as 'PH', 'TH', and 'CH'.

The Voiceless Stops, 'P', 'T', 'C', 'K', and 'Q':

These should always be true quiescent surds in Latin and the plosive allophones so frequently substituted for them in English are best avoided.

'P' is a bilabial, just as in English, but some attempt should be made to always use the quiescent allophone heard 'spell', and avoid the plosive allophone typically heard in 'pelt'. This is a good time to remind readers accustomed to the laxities of English that there are no silent letters in Latin. 'P' is always to be pronounced in all positions.

'T', at least in its primary allophone, is a true dental stop in Latin, pronounced with the tongue actually touching the upper teeth in much the same position as English uses to articulate 'th' as in 'think', but with the breath totally obstructed. It does, however have a secondary allophone when it follows 's' which is indeed a postdental stop exactly like the English sound of 't' in 'stop'. A palletised allophone of 't', which actually sounds more like a palletised 'ts' combination, is often, though not always, heard in Ecclesiastical Latin when 't' is followed by 'i' and then another vowel. One might suppose that this would cause confusion in spelling, but true cases of a 'ts' combination are really quite rare in Latin, as all original occurrences of 'ts' were reduced to simple 's' during the Old Latin period. The distinction between 't' and 'd' is always absolute in Latin, which means that there is no voiced allophone of 't' in Latin as is sometimes heard in some substandard varieties of English. It must also be noted that Latin 't' is never 'swallow'd', or pronounced as a glottal stop, as it sometimes is in some other substandard varieties of English.

'C','K', and 'Q' are perhaps best treated together as their histories in Latin are very much intertwined. When the Greeks adopted the Phonetian alphabet 'K' was a voiceless velar stop and 'Q' a voiceless, and probably rounded, postvelar stop, but the letter which was to become 'C' in Latin was a voiced velar stop which the Greeks called 'Gamma' and wrote as 'Γ'. This would have served Latin quite as well as it served Greek -- better in fact, because the Greeks had no real need to distinguish 'K' (Kappa) from 'Q' (Qoppa) as they were mutually exclusive positional allophones of what the Greeks heard as the same phoneme, while Latin, when its spellings were finally sorted out, would find a real use for the latter letter which the Greeks rarely used except as a counting symbol. The Romans, however, did not acquire the Alphabet directly from the Greeks, but from the Etruscans, a people who used a very mysterious, non-Aryan language, which apparently had a very different phonetic system, and thus very different alphabetical needs. Little as we can know about their language, it appears likely (largely from the evidence of Latin) that the Etruscans may have used no voiced consonants at all in their language, or at least no voiced stops, and that even if voiced consonants were not totally foreign to them, the voiced/unvoiced pairs were for them no more than positional allophones of each other, and not separate phonemes. The Greek letters 'Π' (which they reshaped into 'P') and 'B' were for them no more than two letters doing a job that in their language either one could have handled well enough by itself; and likewise for 'T' and 'Δ' (which they reshaped into 'D'). In fact although the Etruscans kept the voiced stops 'B' and 'D' in their alphabet, they seem to have made little or no use of them in the actual writing of their language. Apparently, however, the Etruscans could not afford to waste 'Γ', (which they would reshape as 'C') as an equally ignored voiced allophone of 'K' or 'Q'; because in their language the prevelar (or postpalletal), velar, and post velar stops were more than the positional allophones they appear to be in most languages, but had evolved into three totally separate phonemes needing three separate written representations to distinguish them from each other, even though each within itself needed no written distinction for any voiced allophone it may have had. Students of Etruscan epigraphy have noted that Etruscan words tended to contain only one written vowel, which appears either initially or after only one or two consonants, but which was often followed by a fairly long series of apparently unpronounceable consonants. Some have deduced from this that the Etruscans regularly placed a very strong stress accent on the first syllable of each word, and that this resulted in the reduction of the vowels in subsequent syllables to brief whispers between unvoiced consonants, which would not have been represented by any written characters of their own. Such a very strong stress accent would also result in the closing of any vowels of medium closeness, and explain why the Etruscans seem to have confused 'E' with 'I' and 'O'(which they almost never actually used) with 'U'. The briefly whisper'd vowels in unaccented syllables were likely to have been neutral after most consonants. But the use of 'C' would have indicated that any unwritten whisper'd vowel which were to follow were to be a close front vowel, and 'Q' would have likewise inply'd that a close, rounded, back vowel or semivowel were to follow in reduced and probably whisper'd form.

As well as this phonemic reassignment may have served the Etruscans, it served the Romans, whose phonetic system was essentially Aryan, and thus in fact much more similar to that of the Greeks, rather poorly. With no other model than the Etruscan, however, Old Latin, for which the 'C', 'K', 'Q' of Etruscan were really only positionally determined allophones of the same phoneme, attempted to use them as they heard them, placing 'C' before 'E' and 'I', 'K' before 'A', and 'Q' before the rounded back vowels 'O' and 'V'. Fellow Aryans as they were, they would of course have been better served had they inherited the Greek alphabet directly, and thus inherited the distinction between voiced 'Γ' and unvoiced 'K'. When the Etruscan influence began to wane, however, and the Romans gained an increasing knowledge and respect for the culture and writings of the Greeks, they saw a way to make their own alphabet serve them much better with only a tiny amendment. They attached a small reversed representation of the Greek 'Γ' to their own letter 'C' to represent a voiced equivalent of 'C', 'K', or 'Q', which for them had always been as much a separate phoneme as it was for the Greeks. Thus 'G' was born, and 'C', 'K', and 'Q' could continue sorting themselves out with the understanding that they would henceforth only represent their original unvoiced sounds. As Latin entered its Classical period some two hundred years before Christ, 'C' began to replace 'K' and 'Q' in almost all positions. 'K' fell into disuse except in some Greek borrowings and a few traditional abbreviations. 'Q', however, having been relieved of the responsibility of representing any unvoiced velar stop that happened to be accidentally rounded by a subsequent rounded vowel, survived in a narrower usage wherein it represented a genuinely separate phoneme inherited from the Aryan parent language. 'C' would take over for 'Q' whenever 'O' followed or whenever a following 'V' was to be sounded as a full vowel, while 'Q' would now be reserved to indicate that a following 'V' was to be sounded only as a semivowel. The forms of the relative and interrogative pronouns were undergoing developments in early Classical Latin which made this distinction very usefull. 'C' in Classical Latin was regarded as a single phoneme which was realised as an unvoiced prevelar stop (like 'k' in 'kin') before 'Æ', 'E', 'Œ', 'I', and 'Y'; and as a velar or postvelar stop (like 'c' in 'cat' or 'cold') in other positions. During the Christian period, however, the 'fronted', prevelar allophone of 'C' evolved into a retroflex postpalletal affricate like 'c' in 'cello', or 'tch' in 'latch', and this allophone of 'C' is now standard in Ecclesiastical Latin before 'Æ', 'E', 'Œ', and 'I'. When this type of 'C' follows an initial 'S' the rather complex series of sounds that one might expect is generally simplify'd to a postpalletal retroflex sibilant like 'sh' in 'ship'. When 'C' is followed by any other vowel or by any consonant it has much the same sound as in English, although, like the other stops, it should not in Latin be exploded. 'K' made a bit of a comeback into Christian Latin in some Greek loan words to take appropriate account of the fact that the Greek 'K' remained a simple prevellar stop, like 'k' in 'kin', before front vowels. This practice was never a consistent one, but the Barefoot Abbess prefers the proper Greek pronunciation of Greek loan words and thus favours 'K' in their spelling when 'C' would encourage an inappropriate sounding. 'Q' is always a voiceless, rounded, postvelar stop, and thus should be sounded as it is in English.

The voiced stops, 'B', 'D', and 'G':

These phonemes are in Latin always very clearly distinguished from their unvoiced counterparts. In a language like Latin which does not emphasize the unvoiced stops be exploding them, it is all the more important to be sure to prevocalise initial voiced stops, which means that the vocal cords begin to vibrate a moment before the breath is released.

'B' is sounded as it is in English.

'D' at least in its primary allophone, is a true dental stop in Latin, pronounced with the tongue actually touching the upper teeth in much the same position as English uses to articulate 'th' as in 'this', but with the breath totally obstructed. It is palletised before the 'i'/'j' semivowel when another full vowel follows, so that 'dj' is pronounced as in 'adjutant'. Note that this last is in effect the same retroflex postpalletal affricate which is used for 'G' when it is followed by any front vowel, and is thus the only real case where the Barefoot Abbess is at a loss to avoid a single, selfsame allophone being shared by two different phonemes, and thus spelt in two different ways. It is well also to note here that in some substyles of Ecclesiastical Latin 'ti' before another full vowel is palletised in exactly the same way as 'c' before a front vowel -- more phonetically symmetrical, perhaps, but less orthographically distinct than what is described above in the discussion of the 't' phoneme.

'G' serves as the voiced counterpart not only of 'C', from which it was derived, but also of 'K' and 'Q'. It thus has more allophones than any other Latin phoneme. Its primary allophones are the voiced velar stop, like 'g' in 'gar', and the voiced postvelar stop like 'g' in 'good'. It needs to be noted here that when 'G' is followed by 'V' and then another vowel it is treated as the voiced counterpart of 'Q' rather than of 'C', and thus the 'V' is always treated as a semivowel rather than a full vowel, and the pronunciation is like 'gu' in 'sanguin'. In classical times there was only one other allophone, a voiced prevelar stop (like 'g' in 'gift') before 'Æ', 'E', 'Œ', 'I', and 'Y'. During the Christian period, however, this 'fronted', prevellar allophone of 'G' evolved into a retroflex postpalletal affricate like 'g' in 'gem', and this allophone of 'G' is now standard in Ecclesiastical Latin before 'Æ', 'E', 'Œ', and 'I'. This is totally appropriate for native Latin words, but the Barefoot Abbess prefers the simple voice prevelar stop when 'G' is followed by 'Y' or by any of the above listed front vowels in words of Greek origin as this answers best to the proper value of the Greek letter 'Γ'(Gamma). It some of the most uncritically Italian substyles of Ecclesiastical pronunciation 'g' combines with a following 'l' or 'n' in such a way that the 'g' is not heard at all as a separate phoneme, but rather causes the following liquid or nasal to be palletised as if it were followed by an invisible 'i' or 'j' semivowel. The Barefoot Abbess does not at all recommend this rather late violation of the historical genius of Latin, but must warn the student that confusing as it may be, it is nonetheless often to be encountered.

Aspirated and Sibilated Stops:

The consonant symbols described so far would have served well enough by themselves to cover the phonemes of native Latin Words, but when the Romans began to absorb Greek culture and borrow words from the Greek language, they had to deal with several new letters which the Greeks thought of as representing discrete phonemes, but which the Romans heard as equivalent to simple voiceless stops followed by 'h' or 's', and thus thought it best to respell them in this way rather than borrow another handfull of Greek letters. To understand this properly, however, it is best to take a closer look at the problems the Greeks had faced in their adaptation of the Phonetian alphabet to the needs of preclassical Greek. It was stated above that the 'h' sound arose in prehistoric Greek as a positional allophone of the 's' phoneme. When they began borrowing the Phonetian alphabet, however, the Greeks represented the sibilant allophone as 'Σ' (Sigma), from the Phonetian letter 'Shin', and used 'H', from the Phonetian letter 'Heth', to represent the aspirate allophone. Eventually, however, they came to reserve the full 'H' to represent a long version of 'E' (which they needed because the distinction between long and short vowels was of greater importance in Greek than it ever was in Latin) and to use only 'c ' (a reduced form of 'H') to represent the aspirate 'h' sound. Even though they did not choose to simply use 'Σ', to represent both sounds, it is still likely that they realised, at least to some extent, that the two sounds, like most pairs of position allophones in most languages, were mutually exclusive -- except that either sound could follow 'p', 't', or 'k' in very intimate combinations which the Greeks heard as single phonemes that were very definitely related, and yet distinct. The Greeks solved this problem by deriving six new letters from three old Phonetian letters. From the Phonetian 'Teth' (a symbol representing a retroflex 't') the Greeks created 'Θ' (Theta) and 'Φ' (Phi) from the phonetian 'Samech' (a symbol for an 's' sound) the Greeks created 'Ξ' (Xi) and 'Χ' (Chi); and from the Phonetian 'Tsade' (a symbol for a 'ts' sound) the Greeks created 'Tsan' (Tsan) and 'Ψ' (Psi). Even as the Greeks were beginning to write, however, their 'ts' phoneme was undergoing the same simplification, at least in most dialects, to a plain 's' sound that befell the equivalent combination in Latin; and thus 'Tsan' (Tsan) became redundant except as a numerical symbol and was generally replaced by 'Σ'(Sigma). (The 'ΤΣ' combination seen in Modern Greek arose only in Byzantine times when 'Τ', 'Π', and 'Κ' began to replace 'Θ', 'Φ', and 'Χ' whenever 'Σ' follow'd.)

The Etruscans appear to have made somewhat of a hash of 'Ξ' (Xi), 'Χ' (Chi), and 'Ψ' (Psi),but they seem to have borrowed 'Θ' (Theta) and 'Φ' (Phi) with little change, and may also have adopted 'Tsan' (Tsan) with something like it's original phonetic value.

The Romans, however, had not only simplify'd 'ts' to simple 's' even earlier than the Greeks had, but had no use for aspirated stops because those sounds had become the continuants 'F', and 'H' in their language, and thus from the whole original collection of unvoiced aspirated and sibilated stops they inherited only 'X', with a phonetic value equivalent to the Greek letter 'Ξ' (Xi), but with more or less the proper position of 'X' (Chi) in the Greek alphabetical order. Willing as they were to borrow the Etruscan alphabet along with a good many Etruscan words, the Romans of that time were still unwilling to burden their language with any foreign phonemes. Their descendants, however, upon discovering Greek culture at the start of the Classical Latin period, were willing not only to borrow a host of Greek words, but took pains to borrow them in a way that shewed the greatest respect for the Greek language.

Aspirated Stops:

Because a direct borrowing of 'Φ' (Phi), 'Θ' (Theta), and 'X' (Chi) into the Latin alphabet would have spelt confusion because the Romans were already long accustomed to using 'X' with the value that the Greeks gave to 'Ξ', and because the Classical Greek phonemes that they represented sounded to the Romans like simple voiceless stops followed by (or perhaps coärticulated with) what the Romans heard as 'h' sounds, the Romans chose instead to use their own familiar letters in the combinations 'PH', 'TH', and 'CH' to represent 'Φ' (Phi), 'Θ' (Theta), and 'X' (Chi) in the Greek words they were borrowing. Students of the Classical style of pronunciation are encouraged to sound these double consonants as in 'top hat' 'felt hat' and 'magic hat', and indeed this is arguably best for reading texts of the Classical period. Three of four centuries later, though, in the Hellenistic Greek that the Latin Fathers was hearing and translating, the 'p','t', or 'c' elements that earlier Romans had heard in the Classical Greek pronunciation of 'Φ' (Phi), 'Θ' (Theta), and 'X' (Chi) would no longer have been heard as full stops, but had become continuants sounding much more like, 'ph' in 'amphibian' (a bilabial continuant, like blowing a fly off the end on one's nose, and not a labiodental continuant like 'f'), 'th' in 'think', and 'ch' in 'loch' but it is likely that the 'h' element, which finally disappeared during the Byzantine Period, was still to be heard, at least in intervocalic positions when the following vowel took the stress accent. These later, Patristic pronunciations are the ones that the Barefoot Abbess recommends for Christian Latin. It is worth noting that even proponents of 'Classical' pronunciation admit that 'PH', 'TH', and 'CH' all produce the best scans in Classical Latin poetry if treated as single consonants, which tends to suggest that they had lost their original stop + aspirate quality and become either simple stops or simple continuants within a generation or so of their adoption into Latin.

Sibilated Stops:

The first thing to be noted about the sibilated stops is that although 'ts' was simplify'd to simple 's' in the preclassical periods of both Greek and Latin, 'x' and 'ps', despite the laxity wherewith they are so often treated in English, must never be so simplify'd in Latin. They are always to be sounded as in 'lax' and 'laps' regardless of their position. They do not, in good Latin, have any voiced allophones. Sibilated stops yield the best scans when treated as double consonants in Classical Latin poetry.

'Z', in both Greek and Latin, is a voiced, postdental, sibilated stop, like 'dz' in 'adz', and is thus in a class by itself. Historically it is quite unique. The Greeks borrowed 'Zain', the seventh letter of the Phonetian alphabet, as 'Z' (Zeta), the seventh letter of their own alphabet, but while 'Zain' was a voiced postdental sibilant, like 'z' in Modern English, the Greeks had little need to distinguish what would for them have been a voiced allophone of 'Σ' (Sigma) that might occur in only a few borrowed words. They did, however, have a real need to distinguish a voiced counterpart of the shortlived 'Tsan' (Tsan) which was not only quite definitely a separate phoneme, as either could occur at the beginning of a word, but was stubbornly remaining a sibilated stop even while 'Tsan' (Tsan) was being simplify'd to a simple sibilant identical to 'Σ' (Sigma). The reason was that this complex voiced sound arose in Greek not only as a voiced allophone of the 'ts' sound for which they might briefly have needed 'Tsan' (Tsan), but also as an uniquely Greek development of the high front semivowel which Latin inherited unchanged from the Proto Aryan parent language. Thus the same Proto Aryan word which became 'iugum' in Classical Latin, and 'yoke' in Modern English became 'ζυγον' in Classical Greek, even though the meaning remained the same in all three daughter languages. When Helenised Jews and Christianised Greeks attempted to bring Hebrew words into the Greek language there were some confusing inconsistencies. 'Z' (Zeta) served most often to transliterate its own parent 'Zain'; but with 'Tsan' (Tsan) already long forgotten except as a numeral, sometimes competed with 'Σ' (Sigma) as a transliteration for 'Tsade'. Those who know Hebrew well enough to sort this out do well in rendering 'tsion' as 'tsion' regardless of whether it is realised in a Latin text as 'sion' or 'zion', and perhaps to pronounce 'Z' as a simple voiced sibilant when it is known to represent the Hebrew 'Zain', but those with less confidence are best advised to always sound 'Z' as 'dz' in 'adz'. 'Z' had held the seventh position in the original version of the Latin alphabet just as it had in the Greek alphabet. Whatever value it had had for the intermediate Etruscans, however, this original Latin 'Z' had probably been used as a simple voiced sibilant, as it represented a brief stage in the transition of the intervocalic allophone of the 's' phoneme into an 'r' sound. When classical Latin writers finally replaced their old 'Z' with 'R', they gave its old place in the alphabet to their newly devised 'G', and added it back to the end of the alphabet as a reminder that it would henceforth be used to represented the more complex sound of 'Z' (Zeta) in borrowed Greek words. Like the unvoiced sibilated stops, 'Z' yields the best scan when treated as a double consonant in Classical Latin poetry.