The Making of Latin/Chapter 1
THE MAKING OF LATIN
I. THE KEY OF THE RIDDLE
§ 1. Etymology means literally ‘telling the truth’; the name was invented by the Greeks to describe the study of what they called the ‘true,’ that is the original, meaning of words. Tor words are continually changing their meaning; and the form, that is the sound, of words is often changed too. For example, the English word quick now means ‘active, speedy’; but in Queen Elizabeth’s time it meant ‘living,’ as in the Biblical phrase the quick and the dead; and in far earlier times, as we shall see, its form was something like gwīgwos (§ 60 (4)). Now people once used to suppose that such changes came quite by haphazard; but we know now that in each language at any given time the different sounds were being changed in certain definite ways, and in no others. By finding out these ways of change we can often discover the earlier form of a word, and its earlier meaning also; and when we have done this, we say we have found its Derivation.
§ 2. School-boys and others often wonder what means there are of finding out the derivations of words. They can only be found when beside the word which we wish to explain there exists some other word or words either in the same language or in another kindred language, with which the word can be compared. By comparing two or more kindred words we can generally find out something about the earlier form of each of them. When we compare Lat. gero, haurio with their participles gestus, haustus (and when we have learnt that in Latin -s- between vowels became -r-), we see that their earlier forms were *geso,[1] *hausio; so if we set temere ‘rashly’ beside Sanskrit tamas ‘darkness,’ when we have learnt that a in Sanskrit regularly corresponds to any one of the three vowels, a, e, or o, in Latin, we see that temere meant originally ‘in the dark.’
§ 3. By Kindred Languages we mean those which have descended from a common ancestor. Thus what are called Romance languages, Corsican, Sardinian, Spanish, Portuguese, French, Italian and Roumanian, are kindred languages because they have all descended from Latin. But they did not all begin their separate crowth at the same date. The oldest are Corsican and Sardinian, which took their rise from the Latin of the Roman camps which were established in those islands between the First and Second Punic Wars in 231 B.C. Then comes Spanish: for Spain became a Roman Province at the end of the Second Punic War, in 197 B.C. Then France, of which the oldest dialect (Provencal) dates from the establishment of the provincia Narbonensis in 121 B.C. Latest of all began Roumanian which was started by the conquest of Dacia by the Emperor Trajan: this was completed in 107 A.D. and was commemorated by the great Column of Trajan which is still standing in Rome.
The relation between Latin and all these languages is sometimes expressed by using the word proethnic which means ‘before the separation of the nations’; Latin is ‘pro-ethnic Romance.’
§ 4. In just the same way nearly all the languages of Europe, and some of the languages of India and Persia, have descended from a language which we call Indo-European. But unluckily we have no direct record of this language: we can only infer what it was by comparing the different branches of language which sprang from it, Indo-Persian, Armenian, Greek, Italic, Keltic, Germanic and Balto-Slavonic (p. 127).
§ 5. The oldest branch of Indo-Persian is Sanskrit, the sacred language of the Brahmins of India; and their most sacred document is the Rig-Veda or ‘Book of Hymns,’ the oldest parts of which are judged to be as old as 1500 B.C.
§ 6. By Greek we generally mean Attic, the language of ancient Athens: but the other dialects, Tonic, Doric, and Aeolic, often had more primitive forms. Attic spread over all the coasts of the Mediterranean, and by the conquests of Alexander the Great, who died in 328 B.C., over the whole of Asia Minor and even further Hast. This wide-spread language was called the Koinē, that is ‘the common dialect’; and from this is descended Modern Greek.
§ 7. Keltic includes in one branch Gallic, an ancient language of Gaul, of which we have only scanty records, and also Welsh and Breton: this branch is called ‘Brythonic.’ In the other branch, which is called ‘Goidelic,’ are Manx, Irish, and the Gaelic of the Scotch Highlands. Breton, Welsh, Irish and Gaelic are still spoken languages.
§ 8. The earliest Germanic tongue known to us is Gothic, recorded in the translation of the New Testament made by Bishop Ulfila, a missionary to the Goths of the Crimea in the 4th century a.p.; other branches are Icelandic, Norwegian, Danish, English (of which ‘Scotch’ is a dialect), Dutch, and German.
§ 9. Latin is the best known language of the Italic Branch, in which are included besides Latin several dialects, called Oscan, Umbrian, Volscian, Sabine and Faliscan, which are known to us from inscriptions and from scattered statements of grammarians. It is often possible to discover the original form and meaning of a Latin word by comparing it with the kindred word in one of these dialects. For instance, North Osc. prismu ‘first’ shows that Lat. prīmus has lost an -s- before its -m- (§ 198) and contains the stem pris- that we see in pris-cus ‘primitive’ (§ 252).
Changes in Language
§ 10. From the examples already given it will be clear that by comparing kindred languages we often discover changes of sound that have taken place in the growth of each. These changes are of two kinds which it is very important to distinguish.
The first kind of change happens unconsciously without the speakers of the language knowing that any change has happened at all. Thus Lat. rēgem became in French roi: and we find that the same change happened to Lat. ē in all words in which it stood in an accented syllable. The statement of any such change (in this case that “Lat. ē when accented became oi in French”’) is called a Phonetic Law. The loss of the -g- and of the -em of rēgem happened also through phonetic changes, or, as we say, ‘by Phonetic Laws.’
§ 11. Strictly speaking Phonetic Laws have no exceptions: that is to say, a phonetic change affects equally the same sound in all words existing in the language at the time, provided that the sound is under the same conditions in all the words. Thus Lat. -ē- became Fr. -oi- when the e was accented, as it was in Lat. rēgem (see § 85 (2)), but not otherwise. Thus Lat. mē, tē, sē, became Fr. moi, toi, soi, when they were emphatic and therefore accented, but Fr. me, te, se, when they were unaccented.
§ 12. But it often happens that words containing the sound affected by a particular phonetic change are introduced into a language (sometimes from being borrowed from another language) after the phonetic change has ceased or, as it is often stated, ‘after the Phonetic Law has expired.’
For example, the word kirk was introduced (or re-introduced) into English from Scotch many centuries alter the change of k to ch in English, which produced the English word church from the same original word, which in Latin was cyriaca (Gr. κυριακή ‘the house of the Lord’), Anglo-Saxon cirice.
§ 13. In that case in modern English we have what is called a Doublet—two words of different form, derived from the same original and existing side by side, often with a slight difference of meaning which gives the key to their history. And it is generally the borrowed word, in this case kirk, which has the narrower and more precise meaning. Now in Latin a great many words were borrowed from the neighbouring dialects, especially Oscan and Sabine, and many from Greek: and many of them have a shape which is different from what they would have had if they had existed in Latin all along. For example, rufus ‘red-haired’ is a borrowed word (§ 177) and shows a narrower meaning than the pure Latin ruber ‘red’ which is closely related to it. We saw in § 11 another type of Doublet (Fr. moi and me) due to different conditions of Accent.
§ 14. But there is another kind of change which is, generally at least, partly conscious, that is to say, present to the minds of many speakers when the change is made. ‘This is due to the influence of Analogy, which is the great constructive and reconstructive force in human speech. How it operates can best be shown by a few examples. We know from kindred words that the Latin numeral corresponding to Eng. nine was to start with *noven (hence nōnus ‘ninth’ from *novenos); but it was changed to novem so as to match septem and decem in which -m was original (cf. Lat. septimus, decimus: Sansk. saptamas, ‘seventh’; daçamas, ‘tenth’),
§ 15. Very often these Analogical changes modify the results of a Phonetic Law and so produce apparent exceptions to it. In Latin in words of two syllables (‘Disyllables’) whose first syllable was short like via and modus, the second syllable of the Ablative was shortened by a phonetic change (§§ 95, 96). But in words like dominus and mensa which had more than two syllables, or a long first syllable, this phonetic change did not take place, so that the abl. remained mensā, dominō: and these words were far more numerous than the words of the shape of via, modus.
Hence people felt that in the First and Second Declensions the meaning of the abl. required -ā or -ō to express it properly: and so they would not allow themselves or their children to pronounce anything but a long vowel in the Abl., even of words like via or modus. We state this process shortly by saying that the -d in via and the -ō in modō were “restored by Analogy.”
§ 16. But meanwhile the word modo had come to be used not only as an ablative meaning ‘by’ (or ‘on’) ‘the proper line’ but as an adverb meaning ‘just, only’: and in this meaning it was no longer felt by the general body of speakers, or as we say ‘in the popular consciousness,’ to be an ablative. Hence in the adverb the short -o produced by phonetic change was left unaltered. Exactly the same phonetic change affected the nominative and accusative neut. plur. of disyllabic words, such as bĕna, ĕa, in which we know from kindred languages that the -a was originally long.
§ 17. But in the compound word intereă ‘among those things, meanwhile,’ which had become an adverb, the original long -a remained. It was not affected by the phonetic change, because the word had four syllables; and as it had come to be felt merely as an adverb, no longer as a preposition and a pronoun, it was not affected by the analogical change which spread the short -a of the very common disyllabic words ĕă, bonă, mălă, mĕd, tŭă, sŭă through all the neuter plurals of pronouns (like illa), nouns (like dōna), and adjectives (like multa). Words like intereā and modo which became separated in consciousness from the sets of forms (in a Declension or Conjugation) to which they once belonged are said to have been “isolated” or “crystallised.” They are often a great help to our study, because in these we have true phonetic forms preserved of which otherwise we might know nothing.
§ 18. The examples just given show the power of Analogy both to restore and to innovate. In Ablatives like via the old quantity was put back on the pattern of words like mensā (§ 15); but in all the Neut. Plurals of the Second Declension the final -a, originally long, has been shortened to match that of the disyllables like ĕă (§ 17).
§ 19. Such modification by Analogy of what we may call the natural form which words have taken by unconscious Phonetic change is particularly common in Compound words. By a phonetic change (§ 122) the final -o- of the first half of compounds like *agro-cola ‘tiller of the fields’ became -ĭ- so that we know the word only as agri-cola.
Thus a type was established for the great mass of compounds, so that we get words like foedi-fragus, ‘treaty-breaking’ from foedus, foederis ‘treaty,’ where logically the first part ought to have been foeder-. Just in the same way in Greek in the very numerous compounds like μῡθολόγος, ‘a teller of ancient stories’ and μῡθολογίᾱ ‘the knowledge of stories, mythology,’ or ἀστρολογίᾱ ‘the knowledge of the stars,’ hence ‘astrology,’ the final -o- of the first part of the word came to be felt as part of the ending; hence new words were made like φυσιολογίᾱ ‘physiology’ from φύσις ‘native power, nature,’ instead of *φυσιλογίᾱ, which would have been correct. Hence in English came the names for the different sciences which are sometimes called the “-ologies.”
§ 20. So it is common in compound verbs to have the form of the simple verb restored. For example in Latin dēlēnio is restored on the pattern of the simple lēnio, instead of dēlīnio which would have been the true phonetic form (§ 129). Such restored forms are called Re-formates.
§ 21. Again by a regular phonetic change (§ 184) claudo became in compounds clūdo (concludo, exclūdo, inclūdo, reclūdo); but these words were so much commoner than the simple claudo that before the time of Juvenal the form claudo had died out altogether, and been replaced by clūdo, which was taken from the compounds. So plicāre instead of *plecere (Gr. πλέκω) or of *plocāre, from expliādre, complicāre. This process is sometimes called Decomposition.
§ 22. The same kind of reasoning by Analogy often conceals the true history of a word. From the negative adjectival compounds in-decēns, ‘unbecoming,’ improbus ‘not excellent, bad,’ were created the verbs indecet ‘it mis-becomes, is unfitted,’ improbo ‘I count bad, I disapprove,’ which came to be felt as compounds of decet and probo. From the adjective operātus ‘full of work, seriously occupied,’ which was derived from opera ‘work’ as ansātus ‘handled’ from ansa ‘handle,’ but which looked very much like a participle (such as hortātus, pālātus) there was formed later on the Deponent verb operāri ‘to devote oneself to (some duty).’ Such new words are called Retroformates.
§ 23. It is to be noted carefully that all the changes which have been described are changes of Sound, taking place in the spoken language, often long before the epoch in which the language was written down. When a language comes to be written down, and fixed spelling is established,—a thing which has happened to the languages of all civilised peoples,—and still more when the language comes to be printed, we generally find that the sounds actually existing in speech are only roughly represented in writing or printing. This is particularly true of English. The spelling of Latin was much more exact; but in all languages which have come to be written, when a spelling is once established, it tends to remain although the actual sound of the word may have changed. For example, the Latin word which about 200 B.C. was pronounced and written deicō (with its first syllable like Eng. day) was frequently still so written in the time of Cicero, when it had long been pronounced dīcō (with its first syllable like Eng. Dee). For this and other reasons the history of the alphabet or.alphabets used to write a language is something quite different from, although sometimes connected with, the history of the language.
Note i
For example, the sign which in the Latin and English alphabet represents the sound f, namely 
, had once, in all the various Greek alphabets, represented a different sound, that of Eng. w. If we ask how the sign came to change its value, we find that the Etruscans, when in the 7th or 8th Century B.C. they took a Greek alphabet to write their own language which possessed a sound very much like Lat. f, found no sign in the Greek alphabet for that sound. The best they could do was to combine two of the Greek signs and 
(the older form of H) which together properly denoted the sound of Eng. wh (pronounced fully and truly as it still is by the educated class in Edinburgh). This makeshift way of writing the sound f appears not merely in Etruscan inscriptions, but in the oldest of all the Latin inscriptions we possess, on the golden brooch found in 1871 in a tomb at Praeneste which archaeologists ascribe to about 600 B.C. The inscription runs thus:
manios med fhe-fhaked numasioi
This would be in classical Latin ‘Manius me fecit Numerio’ (on the Dative form see § 118 Note; on the Perfect §§ 297 and 71). But the Romans, as by degrees they came to decide on an alphabet for Latin, with the help of their neighbours (who at that epoch were also their rulers), the Etruscans and of their somewhat more distant neighbours the Greeks of Cumæ in Campania, did not use , as the Greeks and Etruscans did, to mean w; so they saw no reason for the cumbrous , and wrote simply to mean f.
All this story of the sign is important for the study of the beginnings of civilisation at Rome. But it tells us very little, if anything, that we did not know about Latin as a language; though of course, if we had not known what the sound of in Latin was, we should welcome the evidence of the spelling wh as showing that it was a sound ‘with an h in it,’ ie. (§§ 25, 52) that it was Breathed. And for the study of Etruscan, of which we still know only a little, the evidence is important.
This example will serve to show that the history of the signs of the alphabet in which a language is written is something quite distinct from the history of the language; and it is the history of the language which governs Etymology, and with which this book is concerned.
Note ii
The body of principles which have been briefly explained here (§§ 10-23) is the work of what used to be called the ‘new school’ of Philologists of whom Karl Brugmann of Leipzig who died in 1919 was the greatest. They made Philology an exact science, because their discovery of the invariability of Phonetic Law made it for the first time possible to apply strict tests to any proposed derivation. We ask now, does the suggested derivation accord with the known Phonetic Laws of the language concerned? Or, if it seems to make an exception to any one of them, can the reason for such an exception be found, for instance, in the influence of Analogy? Or, if it involves our recognising a new Phonetic Law, are there any forms which would contradict the new Law and for which no explanation can be found? For example, it was once supposed that the forms of the Latin Passive contained the word sē; amor ‘I am loved’ was derived from “amo sē.” Not to mention a host of other obvious objections we should have to assume that the final -ē of the supposed “amosē” was lost after the -s- had become -r-. But there are a multitude of forms of precisely the same shape in Latin, i.e. forms which end with two long syllables, the last being a long vowel (e.g. amantī, sevērē) which show that a long vowel was preserved, not lost, in such a position. Hence we can say quite certainly that for this reason, even if there were no other, the supposed derivation cannot be the true one. (On the real origin of the Passive see § 311.)
The first result of the recognition of the strictness of Phonetic Laws was to destroy a great number of derivations which had been long taught and of which many still linger in the dictionaries. But the greater knowledge which scientific methods of study have produced, has given us a far larger number of trustworthy derivations than those which it obliged us to discard.
Even the brief summary of the Phonetic Laws of Latin which this Introduction gives will enable us to apply serious tests to any proposed derivation of a Latin word and, generally, to say whether the changes of sound which it would lead us to assume did or did not take place in Latin at the date suggested. And we shall further realise the kind of questions that must be asked and answered before we can be sure of the truth of a proposed derivation in any language. So far as Greek and English are concerned, we shall have to notice a number of their most important Phonetic Laws in the course of our study of Latin.
- ↑ Observe this convenient use of the asterisk. We attach * to forms which are not actually on record but whose former existence we can infer from other forms which do appear in literature or inscriptions.