A Short Historical Latin Grammar: Chapter II: Pronunciation, Accentuation, and Change of Sound

Английский Грамматика Продвинутый 19 век W. M. Lindsay

Pages 8–36 of the book.

Vowels

The rule for the pronunciation of the Latin Vowels is simple enough. Give the vowels the same sound as in Italian, letting the long vowels have what is called the 'close' sound, the short vowels the 'open.' The 'open' sound is precisely what is given to short vowels in English; e. g. the e of ' led' is open e, and Latin sed should be pronounced with exactly the same vowel ; the o of 'not' is open o, and so was the o of Lat. nota. The short Latin vowels accordingly offer little difficulty to English lips. But though the 'close' sound is in English, as it was in Latin, associated with the long vowels, our long vowels, e.g. the long e-vowel [1] in 'they,' 'say', 'fate,' 'fail,' the long o-vowel in 'know,' 'no,' are really not simple vowels but diphthongs; for 'they,' 'say,' 'fate,' 'fail,' are in reality pronounced with the close e-sound followed by y, 'know,' 'no,' with the close o-sound, followed by w.

[1] By 'the e-vowel' is meant the Continental e-vowel, the sound of Germ. 'See' (cf. Engl. 'say'),not of Engl. 'see,' which has rather the 'i-vowel' (Germ. 'sie,' Fr. 'si ') ; by the 'a-vowel' the Continental a-vowel of Germ. 'sah,' Engl. 'father,' not of Engl. 'say.'

So, if we would pronounce the long vowels of Latin with exactness, we must take care to give them the simple sounds of the French or Italian vowels, pronouncing, for example, the e of sē-dēs like the e-sound of Fr. 'été,' and not like the ey- or ei-sound of Engl. 'say.') Latin a had however the same quality, when long and when short, the only difference being in quantity. The a of pater is the same sound as the a of mater, only uttered more rapidly. Here too, if we aim at exactness, we must not give the short Latin a the same sound as our 'short a-vowel,' though we may give the long Latin a the sound of our a in 'father.' For our 'short a' is really a different sound from our 'long a,' standing in fact halfway between ǎ and ě as we see, if we compare our pronunciation of words like 'man,' ' hat' with the German of ' Mann,' ' er hat.' The German vowel is the same as the a of Engl. 'father' or Germ. 'Vater,' while our 'man,' 'hat,' 'bat' have in them something of the sound of 'men,' ' bet.' Latin pater then should not be pronounced like our 'patter,' [1] but with the a-sound of 'father' rapidly uttered. Another exception to this rule of the different quality of long and short Latin vowels is furnished by the -it of the 3 Sg. Pres. Ind. of the Fourth Conjugation. This i was a long vowel till the second century B.C., when it was shortened owing to the difficulty found by the Romans in keeping up the long sound of a vowel before a final -t (§16).

[1] This was a Roman 'cockneyism' of the fourth or fifth cent. A.D., for we hear of a mispronunciation at that time by the lower classes at Rome of stdtim so that it sounded like stātim, so that it sounded like stětim.

But this -it, when it became a short syllable, retained the quality, though not the quantity, of long i, so that the -it of, let us say prodit, from prodeo, or vincit from vincio, or condit from condio, had the close i-sound (Fr. si, Ital. si), unlike the -it of prodit, vincit, condit from prodo, vinco, condo, which had the open i-sound of our 'it,' 'bit.' And it is probable that some other short Latin vowels also, which were originally long, continued to retain the close sound of the long vowel (see §16). In dies, where it stood before a vowel, i had the close sound, like our sound of 'the' before an initial vowel. The short i of optimus, which took the place of u in the spelling of Julius Caesar's time, had a sound between -u- and -i-, the sound of German modified u (written ü) or the u of Fr. lune; and a preceding v seems to have given any short i this sound in the time of the Empire, e. g. vir, virgo, virtus [1]. Here is a scheme of the pronunciation of the Latin vowels:

Ā Engl. a in 'father.' Ǎ the same more rapidly uttered. Ē Fr. e in 'été.' Ě Engl. e in 'led.' Ī Fr. i of 'fini.' Ǐ Engl. i of 'in.' Ō Fr. au in 'chaud.' Ǒ Engl. o in 'not.' Ū Germ. u in 'gut.' (Plautus compares the repetition of the pron. tu to the hooting of an owl. The y-sound which we insert before u in 'tune,' &c. was unknown in Latin.) Ǔ Engl. u in 'full,' oo in 'good.' (What we often call 'short u,' e. g. the vowel-sound in 'but,' is not a u-sound at all, but is properly called ' the obscure vowel.')

[1] The sound of a Latin vowel was often affected by a neighbouring consonant. Thus jājunus, the older form, became jejunus, the a having assumed the sound of e (open e) under the influence of the repeated j (cf. Vulg. Lat. Jenuarius for Januarius, Ital. Gennaio, with open e); o took the close sound (a sound approaching to u) before rn and other consonant-groups, such as nd (cf. the spellings turnus for tornus, frundes for frondes) ; e took the close sound (a sound approaching to i) before rg (cf. the spelling Virgilius for Vergilius).

Care should be taken to give long vowels their correct sound, both as regards quantity and quality, in whatever position they stand in the word. Thus the i of nobis must be pronounced with the same length and with the same close sound as the i of si; the e of legem must be the same as the e of legis; illegem, 'lawless,' must be pronounced differently from illegem, ' alluring' ; non must have the o-sound of nonus; bellus and stella should differ as Ital. 'hello,' which has open e, and 'stella,' which has close e, and this close e should be pronounced long ; similarly molle and corolla (from corona), and so on. The long sound was always given to a vowel before ns, nf. Thus the preposition in, pronounced like our 'in,' became in- (more like our 'e'en') in compounds like injicio, insilio ; con- (Engl. 'con') became con- (more like Engl. 'cone') in consul, &c.; and the n eventually disappeared in pronunciation, as it disappeared in -O-. Engl. fif 'five' (Germ. fünf), gōs 'goose' (Germ. Gans). It will be found that the rhythm of Latin poetry will be much improved by giving the Latin vowels their proper sounds, instead of the English sounds, as we may see, if we reflect how a French or Italian line would suffer if we were to pronounce the vowels in it like English vowels. But the full appreciation of Latin poetical rhythm can only be attained, when we learn to assign the proper difference of duration to the long and short vowels, by dwelling, for example, on the first syllable of mater double the time that the voice rested on the first syllable of pater.

The rule for the pronunciation of Latin Diphthongs is also an easy one: – Give both vowels of the diphthong their own vowel-sounds, and combine the two sounds in one syllable. Thus the diphthong of neu, the shortened form of nē-vě, combines in one syllable the e-sound and the u-sound: ain (a monosyllable) of ain vero? has the a-sound and the i-sound of disyllabic ais in close combination. The diphthong -ae- had originally the ordinary a-sound followed by the e-sound, just as -ae- is pronounced in Welsh (e. g. maen 'a stone'), although by the end of the Republic and the beginning of the Empire the first part must have begun to degenerate into the sound of our 'short a' in 'man,' 'bat,' a sound which, as we saw, approaches the open e-sound of ' men,' ' bet'; ' for at a later period [1] we find ae pronounced as a simple vowel exactly like a lengthened open e; the first syllable of aequus, just, and equus, a horse, differing merely in duration of utterance, and in no other respect. Oe, which is found in a few legal and poetic, words like foedus, amoenus, as well as the Greek loan-word poena (ch. x. § 11), must have had its first element similarly affected, the o having the sound of German modified o (written ö), for we find oe sometimes used to express the long Greek υ, which had a sound like German modified -u-. The a of au was influenced too by the accompanying u in the direction of the sound of our a in 'water,' ' wall,' with the result that au ultimately reached the sound of long o (open o). Thus the first element in these diphthongs was affected by the second in the same way as we have seen the Latin vowels to be affected by a neighbouring consonant (p. 10 n.).

[1] The pronunciation of ae as e (open e) and of au as o (open o) was a feature of country dialects (cf. plostrum, olla) much earlier. Cicero himself in his letters often uses the more homely forms with o, e.g. loreola for laureola, and his rival Clodius was the first of the gens to change the name Claudius to its plebeian form Clodius, with the view of conciliating the mob. There is a story of Vespasian being reproved by a certain Florus for using the pronunciation plostrum and turning the tables on his critic by addressing him as 'Flaurus.' Cicero's contemporary, Varro, found this 'rustic' ae a convenient expression of the open e-sound of Gk. η in σκηνή, &c., which he wrote scaena; but an attempt to change the name Caecilius to the rustic form Cecilius (like Claudius to Clodius) was successfully laughed down by Lucilius in the century before (Lucil. ix. 10 M.): Cecilius pretor, ne rusticu' fiat.

Scheme of Latin Diphthongs

AE Pronounce with the vowel of 'man' rapidly followed by the e-vowel. AU Pronounce with the vowel of 'water' rapidly followed by the u-vowel. EU as in Ital. 'neutro' (-e- followed by -u-). OE Pronounce with Germ. -ö- rapidly followed by the e-vowel. UI as in Ital. 'colui ' (-u- followed by -i-).

The Latin Consonants do not offer much difficulty. We may give b, p, d, t, g, c the English pronunciation, being careful however always to give the two last the 'hard' sound which we are in the habit of giving to Greek γ and κ, and never the 'soft' (palatalized) sound which the letters have in English words before vowels like e, i. We must pronounce the c of civitas, as of cavitas, and the g of agito, as of ago, like the c of our 'cavity' and the g of our 'ago', not like the c of our 'city' and the g of our 'agitate', 'age'.

-V- had the sound of our -w-, -j- the sound of our -y-; and if we give these letters their proper sound, we shall not have much difficulty in seeing the etymological relation of nē-ve and neu (with the finale dropt, §12), cav-i-tum and cautum, jam and etiam, etc., nor in understanding how the caw of a crow could be mistaken for a cry of avē, avē {Phaedrus). Between vowels [1] -j- was doubled in pronunciation; ejus, for example, was pronounced 'ey-yus,' (with open e) and was often written EIIVS (p. 7 n.). R, called the 'littera canina,' because its sound resembled the growl of a dog (cf. hirrio, to growl), must be given the trilled sound of French and Italian -r-.

[1] -V- was often dropped between vowels, especially similar vowels; e. g. lavābrum (the old form) became lābrum, a bath, si vis became sīs, 'if you please'. Cf. our 'Hawarden' pronounced with loss of w.

Final -m- should not be sounded, but the preceding vowel should be pronounced as a nasal vowel, so that, e. g., -om- will sound like Fr. 'on.' In poetry when the next word begins with a vowel the a of, e. g. illam igitur is elided like the a of illa igitur, the only difference being that the a of illam is nasal. Before g the letter n (as in angulus) had the same sound as our n in 'angle.' S must always have the 'hard' sound that it has in our noun 'use,' never the 'soft' sound of our verb 'to use' ; so pronounce usus like our noun, not like our verb.

Scheme of Latin Consonants

B as Engl. C as Engl. k. D as Engl. F as Engl. G as Engl. g in 'ago,' not as g in 'age.' H as Engl. J as Engl. y. K as Engl. L as Engl. M as Engl., but when final Lat. -m should be dropped and a nasal pronunciation given to the preceding vowel, e. g. Lat. -om like Fr. on. N as Engl. P as Engl. Q as Engl. R as Scotch or Continental r, stronger than r in 'opera.' S as Engl. s of the noun 'use,' never as s of the verb 'use.' T as Engl. V as Engl. w. X as Engl.

To these we may add the Greek letters, y, z, th, ph, ch, which, as we have seen, occur only in Greek loan-words (ch. i. § 6). -Y- (Greek Upsilon) had the same modified u-sound as the -i- of optimus; -z- had the soft s-sound of our verb 'to use'; th, ph, ch, were pronounced as in our 'ant-heap,' 'up-hill,' 'ink-horn.'

The use of the sign -v-, which suggests to us a sound like that of Engl. v for Lat. vos, &c., is of quite a late date. V and u were not distinguished in Latin inscriptions or early MSS., the discrimination in MSS. of the capital form V for consonantal u, and of the Uncial form U for vocalic u being of a late date. Even now we generally print the texts of the older Latin writers, Plautus, Terence, &c., with u (and i), not with v (and j), partly to give their language an archaic appearance, but mainly because a large number of words which in the Classical period, or the Empire, had the w- (and y-) sound had in earlier times the sound of the vowel; larua, a spectre, mask, for example, is a trisyllable in Plautus, never a disyllable; cf. Plautine miluos, a kite, relicuos (gratiis), for class. milvus, reliquus (gratis).

The sound of our v (labiodental) is indeed the sound to which Latin v (bilabial) ultimately developed (e. g. Ital. 'vostro' from Lat. voster, has our v-sound, as Ital. 'giurare', from Lat. jurare, our j-sound) and the change probably began with initial v; but whether Cicero's preference of con- to com- in compounds like convocat, convalescit, can be taken as evidence that vocat, valescit, had already in his time a labiodental sound is very doubtful. Intervocalic v certainly retained its bilabial sound till much later, as we see from mispronunciations of the Empire like paimentum for pavimentum [1]. -B-, which from a bilabial mute (our -b-) had between vowels become a bilabial spirant (our -w-), is regularly used for intervocalic -v- in the third century A.D., and it was probably not till the fifth century that Latin v came to be sounded like our -v-. After -r-, the bilabial spirant -v- came in Imperial Latin to be sounded like the bilabial mute -b-, whence spellings like corbus (Fr. corbeau) for corvus, and the classical ferbui, Perf. of ferveo.

[1] There was at all times a tendency to drop -v- before the accent. The conversational form of avunculus in Plautus' time was aunculus (cf. French 'oncle'), a trisyllable.

Greek θ, φ, χ, had been in loan-words expressed by t, p, c in the Latin of the Republic. Plautus puns on Chrysalus (Crusalus) and crucisalus, from crux and salio (Bacch. 362), on Charinus (Carinus) and careo (Pseud. 736), on Thalem Acc. (Talem) and talentum (Capt. 274). We find also -b- for -φ- in old spellings like Ennius' Bruges for Φρύγες, and in ballaena for φάλλαινα, a whale, and the Vulgar Greek pronunciation of -x- as k-kh has left traces of itself in the conversational Latin of Plautus (Acc(h)eruns for Ἀχέρων), and in the form admitted into classical usage, bracc(h)ium for βραχίων. But at the time of Cicero, when the Greek study of Phonetics was introduced into Rome by Tyrannio, the teacher of Cicero's friend Atticus, it was felt necessary to express the Greek Aspirates more accurately by th, ph, ch; and this pronunciation was carefully followed in polite circles. The struggle to attain the new shibboleth of fashion led to ludicrous misapplications of the h-sound by the uneducated classes, which have been satirized by Catullus in his famous epigram on Arrius (84):

Chommoda dicebat, siquando commoda vellet // dicere, et 'hinsidias' Arrius 'insidias'.

Nigidius, a Grammarian of Cicero's time, emphasized the importance of correctness in the use of the letter h: rusticus fit sermo si adspires perperam ; and the dropping of h seems to .have been even in the time of St. Augustine an unpardonable breach of manners (Confess. 1. 18: si contra disciplinam grammaticam sine adspiratione primae syllabae 'ominem' dixerit, displiceat magis hominibus, quam si contra tua praecepta hominem oderit, quum sit 'homo'). Cicero (Orator 48. 160) tells us that he was forced in spite of his convictions to yield so far to popular usage as to pronounce pulcher, Cethegus, triumphus, Kartha,go, though he still adhered to Orcivius, Mato, Oto, Caepio, sepulcrum, corona, lacrima. Quintilian (1. 5. 20) says: diu deinde servatum ne consonantibus [veteres] adspirarent, ut in 'Graccis' et in 'triumpis.' Erupit brevi tempore nimius usus, ut 'choronae,' 'chenturiones,' 'praechones' adhuc quibusdam in inscriptionibus maneant, qua de re Catulli nobile epigramma est. By the fifth century A. D. Greek -φ- had become a spirant, differing from Lat. -f- only in being bilabial, while -f- was labiodental [1]; and from this time onward -f- is the normal equivalent of -φ- in Greek loan-words, e. g. strofa (Gk. στροφή); cf. Ital. filosofia, etc. The difference between the two sounds in Cicero's time is seen from Quintilian's story of Cicero ridiculing a Greek witness who could not pronounce the first letter of the name Fundanius (Quint. i. 4. 14).

[1] Latin -f- was originally bilabial, but became labiodental in the last centuries of the Republic. Its bilabial character is seen from old spellings like im fronte, comfluont (class. confluunt).

Between vowels the omission of -h- was sanctioned by current usage in a number of words, such as nemo (for ne-homo), debeo (de-hibeo), praebeo (*prae-hibeo), praeda (for *prae-hida, from prehendo). By the first century A.D. prendo and nil had established themselves in current usage, also deprendo, though reprehensus was heard as well as reprensus.

The same wave of syncope that passed over later Latin, reducing vetulus to veclus (cf. Ital. vecchio), etc. (§ 13), made Tityus out of Titius, hodye out of hodie, uncya out of uncia, etc. Through this combination of -y- with a preceding consonant in unaccented syllables, a new series of sounds, unknown in Latin, has arisen in Romance. Latin simia, 'a monkey', has become Fr. singe (through simya), Lat. apium, 'parsley', Fr. ache (through apyum), Lat. rabies, Fr. rage (through rabyes), Lat. cambiare, Fr. changer (through cambyare).

-dy- became identified with -gi, ge-, and Latin -j- (our -y-), and has assumed in Italian the sound of our -j-, e. g. Ital. giorno from Lat. diurnus; while -ty- has developed in Italian into the sound of -ts-, a sound reduced in French to an s-sound, in Spanish to a sound like our -th- in 'thin' (written in Spanish z), e. g. Ital. piazza, Fr. place, Span. plaza, all from Lat. platea through platya. The grammarians of the later Empire have fortunately left us a good many remarks on this change of sound, so that we can trace pretty clearly the course of its development in Latin. The palatalization of t seems from their account to have begun in the fourth cent. A.D., and to have been fairly established by the fifth. About the same time -cy- became assibilated; and so confusions of -ci- and -ti- before a vowel are common in late inscriptions and in MSS. (§ 10).

The palatalization of -c- before -e, i-, e. g. decem, was much later than the palatalization of -cy-, e. g. decies, certainly not before the sixth or seventh centuries A. D. For no grammarian hints at a difference of sound in -c- before a broad and before a narrow vowel, although the assibilation of -ti, ci- before a vowel is mentioned again and again. Greek transcriptions of Latin words with c invariably reproduce it by K, in cases like ΚΗΝΣΟΝ for censum, ΚΡΗΣΚΗΝΣ for Crescens; Latin loan-words in Welsh (first to fifth centuries) show that Latin c was hard in all positions, e. g. Welsh cwyr (Lat. cera), ciwdawd (Lat. civitatem), and similarly German Keller (Lat. cellarium), Kiste (Lat. cista), etc. ; it is not till the seventh century that spellings like paze for pace assert themselves on inscriptions. G before -e, i- may have assumed the sound -y- at a somewhat earlier period. But the dropping of g between two vowels in late spellings like vinti for viginti ( the precursor of Ital. venti), trienta for triginta, cannot be dissociated from spellings like frualitas for frugalitas, where it is dropt before a broad vowel, and points merely to intervocalic g having become, when pretonic, a spirant, like -g- of German Tage, just as intervocalic b became a w-sound in the third cent. A. D. (§ 5).

The Latin writers on Grammar tell us that -l- had a 'pinguis' or 'plenus sonus' in two cases:

(1) when it ended a word or when it was followed by another consonant, e. g. sol, silva, albus,

(2) in combinations like fl, cl, e. g. flavus, clarus;

and an ' exilis' or 'tenuis sonus':

(1) at the beginning of a word, e. g. lectus, lana, lupus, and especially

(2) double -l-, e. g. ille, Metellus, Allia.

The development of Lat. -l- in the Romance languages points to post-consonantal l having been pronounced with what phoneticians call an 'off-glide,' e. g. c(l)larus (Ital. chiaro from clyaro), and pre-consonantal -l- with an 'onglide,' e. g. allier (Fr. autre from aultre). I.-Eur. -ě- in Latin became -o- before -l-, e. g. vow, but remained before -ll- and before -l- followed by the vowels -e- or -i-, e. g. velle, velim, which points to -ll, le, li- having had more the palatal l-sound. So in Spanish the sound known as 'l mouillée' is given to Latin -ll-, e. g. 'caballo, 'a horse,' the sound given by Italian to Lat. -li- before a vowel (ly), e. g. miglia (cf. our 'million'), bigliardo (our 'billiards').

The repetition of -l- in a word was avoided by the substitution of -r- for one -l-. The I.-Eur. suffix -tlo- (ch. XI. § 9), Lat. clo-, became cro- after a stem with -l-, e. g. fulcrum from fulcio, sepulcrum from sepelio, lavacrum from lavo, simulacrum from simulo; and the Latin Grammarians prescribed the use of the suffix -ris when the stem contained an -l-, of -lis when it contained an -r-, e. g. auguralis, militaris. The reluctance of the Romans to begin two successive syllables with a consonant followed by -r- is seen in forms like praestTgiae for praestrigiae, legerdemain, increbui for increbrui.

Metathesis of -r- (and -l-) was as common in bad Latin as in bad English, and mispronunciations of the kind are often censured by the Latin Grammarians. We have in Plautus, e. g. Phrygio (Aul. 508) for Phrygio, corcotarii (Aul. 521) for crocotarii. Colurnus is the Adjective from corylus, a hazel.

At the end of the third century B.C. the (dialectal!) substitution of l for d found its way into the literary language in a few words, where it seemed to be sanctioned by etymology. Dingua (cf. Engl. 'tongue') became lingua (by analogy of lingo, I lick), dacruma (cf. Gk. δάκρυ) became lacruma (by analogy of lacer?), dautia, the enter tainment of ambassadors, became lautia (by analogy of lautus), *odeo (cf. Gk. ὄζω) became oleo.

Another temporary pronunciation, which did not however leave a permanent mark on the language, was the substitution of an -r- sound for d before -f- and -v- about the same period. This is seen in old spellings like amma for advena, arfari for adfari, and is preserved in the legal formula scribendo arfuerunt (written SCR. ARF.)

A caution too must be given about the pronunciation of Latin Syllables, which should follow the Italian more than the English fashion. Each syllable should be pronounced distinctly, with its due share of utterance; it should never end in a consonant, if the consonant can possibly be pronounced at the beginning of the next syllable. And double consonants must be pronounced double, as in Italian, with one at the end of the first syllable and the other at the beginning of the second. So pronounce pro-fu-gus, not 'prof-gus,' be-ne, not 'ben-e,' ma-gnus, rather than 'mag-nus,' bucca and penna like our 'book-case,' pen-knife,' and so on [1].

[1] A mispronunciation like 'jusit' for jussit, 'ile' for 'ille' was, we are told, one to which the Greeks were especially liable.

Double s and double t passed in certain circumstances into the lengthened sound, hardly distinguishable in the case of these consonants from the repeated sound, and were reduced to single -s- and single -l-; -ss-, after a diphthong or long vowel, -ll-, after a diphthong. Thus caussa, glossa, 'a difficult word' or 'the explanation of a difficult word' (Gk. γλῶσσα), formossus, paullum, the spellings of Cicero's time, became at the beginning of the Imperial Age causa, glosa, formosus, paulum. So with -ll- between long -i- and another -i-, e. g. milia, Plur. of mille, vilicus, a farm-bailiff, from villa, and with -nn- in conubium, conecto, conitor for connubium, etc.

A double consonant at the end of a word is not found in Latin. Thus we have miles for *miless from the stem milit- with the Nom. Sg. suffix -s (ch. iii. 8); es, thou art, for *ess from the root es-, to be, with the 2 Sg. suffix *-s (ch. vi. § 19); hoc for hocc (cf. hoc-ei-ne) from the Neut. Pron. *hod with the particle c(e) (ch. v. § 3). Before a word beginning with a consonant these would at all times have their double- s, c- reduced to single -s, c- and this pronunciation gradually came into fashion before a vowel also. But in the time of Plautus we find the last syllable of such words always a long syllable, and even in the classical time hoc (and after its example hic, ch. v. § 3) is found as a syllable long by position, pronounced, though not spelt, hocc before an initial vowel, with short o and double e.

'Doublet'-forms like these, *es before an initial consonant, *ess before an initial vowel (cf. our 'a' before a consonant, 'an' before a vowel) are common in 'Latin. On ac (f or *atc) and atque, nec and neque, proin and proinde, see point 12, which was at an early period dropt in pronunciation after a long vowel, passed through this 'doublet' stage, the d-less forms having been originally confined to cases when an initial consonant followed. In the time of Plautus the only relics of-dafter a long vowel are the monosyllables med, ted, sed, haud. They have the d-less form before a consonant, e. g. me tamen! hau scio! and the first three often before a vowel too, e. g. me enim as well as med enim. By the classical time the d-less forms, me, te, se, have driven the d-forms off the field, but the form haud continued to assert itself.

Latin spelling was phonetic, representing the actual sounds uttered in pronouncing the word, the cases where a spelling was due to Grammarians' theories being few and exceptional. Such a 'grammarians' spelling' was the -bs- of urbs, which was pronounced (and often spelt) -ps-, it being impossible to pronounce -b- along with the Latin -s-, which had the hard (unvoiced) sound of our noun 'use' (§3). The spelling with b is due to the analogy of the Gen., Dat., etc., urbis, urbi. A variety of spelling may generally be taken to represent a variety of pronunciation. For example spellings like hospicium for hospitium (from hospes,-itis), concio for contio (from *co(n)uentio), nuncius for nuntius (from *noventius) are spellings not earlier than the fifth cent. A.D., when -ti- and -ci- had both, before a vowel, come to be sounded with an a-sound ( § 7); spellings like hereo for haereo are later than the fourth cent A.D., by which time -ae- and -e- (open -e-) bad become indistinguishable in pronunciation (§2): spellings like autor for auctor are also late and point to the pronunciation of -ct- as -tt- (cf. Ital. otto from Lat. octo). On the other hand optumus, maxumus, etc. are early spellings, before the time of Augustus (see §1); so are the double-consonant forms, caussa, formossus, etc. (see §9); equos, etc., loquontur, etc., are the oldest spellings, then ecus, locuntur, then in Trajan's time equus; loquuntur (see p. 31). The uncertainty about the pronunciation of -h- (§6) is reflected in spellings like arena for harena, olus for holus, vegetable, ariolus for hariolus, a soothsayer, humeo and humidus for umeo and umidus. Often a wrong spelling is due to a perverted etymology; thus deliro was written delero, as if connected with Gk. ληρῶ; letum was written lethum (cf. Engl. lethal) and referred to Gk. λήθη; sepulcrum from sepelio (§8) was fantastically derived from se, without, and pulcer, misspelt pulcher, and was written sepulchrum (on pulcher, sylva, etc. for pulcer, silva see ch. i. § 6; cf. Engl. 'sepulchre,' 'sylvan'); cena appears at a very early time as coena, through confusion with Gk. κοινός; comminus (cum and manus) was wrongly written cominus on the analogy of eminus and so on.

The rules of the Latin Accent are so simple that it was never found necessary to indicate by accent-marks the syllable on which the accent was to fall, as was done in Greek. In Latin the quantity of the penultimate syllable regulates accentuation, as the quantity of the final in Greek. When the penultimate syllable is long, the accent rests upon it; when short, on the antepenultimate, e. g. decōres, décǒres. This is also the natural English practice. There are however a few points of difficulty. Monosyllables with a long vowel, like flos, mos, res, had a 'Circumflex' Accent, the voice rising first and then falling slightly, and so had long final syllables of words whose last vowel has been dropt by Apocope or Syncope, e. g. illic, 'there,' from *illice, nostras, 'of our country,' from nostratis (Plaut.), audit for audivit, adduc for adduce, while monosyllables or apocopated finals with a naturally short vowel had the ordinary 'Acute' Accent, e. g. nix (nivis Gen.), pars, illinc, from *illim-ce. Again the Voc. and Gen. Sg. in -i of Nouns in -ius, -ium, were from the beginning of the Empire, or earlier, pronounced with the accent on the paenultima, even when the paenultima was short, e. g. Valéri, Vergíli, tugúri. This accentuation, introduced by Latin of Sentence writers on Grammar under the mistaken idea that these forms were contractions of an earlier Valérii, Vergílii, tugúrii, etc. (ch. iii. §6), was observed in the speech of the educated classes (cf. nostras). Further some of the usages of what is called 'Sentence-Accentuation' must be noticed, for the accent which a word would bear, if uttered separately, may be different from the accent assigned to it, when standing with other words in a sentence. The Greek Preposition πρός, for example, had, if mentioned by itself, an acute accent. But in the sentence its accent was obscured by the Noun which it governed, e. g. πρὸς πόλιν and this by the Greek system was expressed by replacing its acute with a grave accent. Similarly in Latin a Preposition, say súpra, or a Relative, say quális, were in sentences like supra moenia stat, talis est, qualis Cicero fuit, united with the following word into a word-group which took the ordinary accent of a single word, supra-moénia, qualis-Cícero. The Adverb supra and the Interrogative qualis were not subordinated, but retained their independent accent, súpra habitat, quális fuit Cicero? Other words which became subordinate or enclitic in the ordinary utterance of the sentence were:

(1) the various parts of the Substantive Verb, e. g. amatus-est (often written by the Romans amatust),

(2) the Personal and Possessive Pronouns, unless specially emphasized, e. g. tu-me, ad-me (cf. Gk. πρός με, πρός σε or our 'for him,' 'with him,' with the stress of the voice on 'for;' 'with,' unless the Pronoun is emphasized, as in 'for me, but not for thee'), meus-dóminus, mea-dómina (Ital. Madonna),

(3) the Demonstrative Pronouns, when unemphatic, e. g. ille-dóminus (pronounced with Syncope of -e, ill'-dominus); from this unemphatic use of the Demonstrative Pronoun has sprung the Romance Definite Article, e. g. Ital. 'il padre,' Span. 'el padre,'

(4) Conjunctions like et, sed,

(5) Auxiliary Verbs, such as volo in volo-scíre, cave in cave-fácias. Such Auxiliaries have in the languages derived from Latin been reduced to mere Tense-signs, e. g. Fr. 'j'aimerai' from Lat. amare-ha(be)o, 'j'ai fait' from Lat. ha(be)o-factum.

The subordination of many of these words was indicated by the Roman way of writing; the Preposition was often written-along with its Noun in one word, just as it was along with its Verb in a Compound Verb, incurrum, like incurro (cf. imprimis, obviam, admodum, denuo, for de novo, §15), while Auxiliaries like iubet, volo, were in certain cases similarly treated, e. g., quolibet, quamvis, quantumvis. And spellings like quomodo, quare, postridie (for < *posteri die), decemviri show that these Nouns, modus, res, dies, vir, were subordinated in Latin speech, just as in English 'thing,' 'kind,' 'part,' etc., are used without stress in phrases like 'something (nothing) of that kind,' 'some parts of England.' The Enclitics -que, -ve are examples of I.-Eur. Enclitics (cf. Gk. τε, *ϝε in ἠὲ rom *ἦ-ϝε). We saw above (§9) that the Latin pronunciation, like modern Italian, gave each syllable its due share of utterance, while the English slurs the unaccented syllables, and so has, for example, reduced 'fantasy' to 'fancy' and has given 'minute' (Lat. minutum) the sound of 'minit'. But this difference between Latin and English is not more than one of degree. For the Latin Accent was like ours a stress-accent; at all events it was an accent mainly of stress, though this stress was probably accompanied by a slightly higher tone than the tone of the unstressed syllables. And like all languages with stress-accent Latin had, though not to the same extent as English, the tendency to weaken the unstressed syllable, a tendency still seen in its modern representative, the Italian language (cf. Ital. 'gridare' from Lat. quiritare 'to shout,' 'shout for protection,' 'balsimo' from Lat. balsamum, albero from arborem). The Latin stress-accent left traces of itself at all periods of the language in the Syncope, the Weakening, and the Shortening of unaccented vowels, three processes which have so materially altered the appearance of the language that they must be treated in detail in separate paragraphs.

For a century or two before the literary period the law of accentuation which was mentioned in the last paragraph, the 'Paenultima Law,' as it is called, did not prevail. Every word was at that time accented on the first syllable. Long words had, of course, as they had at all periods of the language, a secondary accent as well as the main accent [1], but in them, as well as in shorter words, the main accent rested on the first syllable.

[1] Cf. any English long word, say 'characteristical,' which has the main accent (´) on the antepenultimate, the secondary (`) on the first syllable.

The change to the Paenultima Law of Accentuation began in these long words with the substitution of the main accent for the secondary, the secondary for the main; e. g. témpestàtibus became tèmpestátibus, élemèntum became èleméntum (like our 'èleméntal'). The change to the Paenultima Law was not wholly completed at the beginning of the literary period, for the accent was still on the first syllable of words like fácilius, bálinĕum (later balneum), that is to say words of the scansion [short-short-short-short/long], in the time of Plautus. Instances of Syncope under the Early Accent Law are:

(1) of Prepositions in Compounds, e.g. anculus, an old word for 'servant' (whence ancilla), for *ambi-quolus (Gk. ἀμφί-πολος). By Syncope of this kind the old Preposition ambĭ was reduced to am- in Compounds ; similarly the old Preposition endŏ, indŏ, or indŭ was reduced to ind-, in-, and became confused with the Preposition in; e.g. induperator came to be pronounced as imperator, indugredi as ingredi;

(2) of the second syllable of the first member of a Compound, e.g. hospes for *hosti-pet-s, from hostis in its old sense of 'a stranger' ; princeps for *prīmi-ceps; forceps, the smith's tongs, from *formi-ceps from formus, hot, connected with Greek θερμός (ch. x. § 18); quindecim from quinque and decem; undecim, from ūnus and decem; vindemia, vintage, for vīnĭ-dēmia, from dēmo;

(3) of the first syllable of the Reduplicated Perfect of a Compound Verb, e. g. rettuli for re-tetuli (ch. vi. §10), repperi for re-peperi, reccĭdi or re-cecidi, where the double consonant preserves a trace of the Syncope.

When a short vowel following r was syncopated, the pronunciation of r without a vowel (as in the final syllable of Fr. 'sacre,' &c.) led to the form er. Thus *sacrŏ-dos, *sacrĭ-dos became sacerdōs; *sacrŏ-lom became *sacerlom, sacellum. Similarly -lĭ- appears in the syllable after the accent as ŭl (older ŏl) in facultas, &c. A syllable like -vĭ / vĕ- became after Syncope u in autŭmo for *ăvĭtumo (cf. Gk. οἴω for ὀϝ-ιω), auceps for *ăvĭ-ceps. But forms like lābrum for the older lăvābrum, dīnus for dīvīnus, audissem for audivissem, and the like, are not due to Syncope, but to the habit of dropping v in pronunciation between two vowels, especially similar vowels. Nor should we refer to Syncope such a change as dixti for dixisti, misse for misisse, debilitare for debilitatare, idolatria for idololatria, where, owing to the unpleasant effect of two neighbouring syllables having the same sound, one has been suppressed. *Saeclum *is not a syncopated form of saeculum, for saeclum is the older, while in saeculum a vowel has been inserted between the c and the l to aid pronunciation, like the parasitic ῐ of mina, a Greek loan-word, µvᾶ, with which we may compare Fr. 'canif,' a German loan-word, 'Knif,' or the parasitic u of Tecumessa (Τέκμησσα), for which Tecmessa was first used in Cicero's time [1]. On this ending -culus, -culum, and -clus, -clum, see ch. xi. § 9.

[1] Other examples of the Parasitic Vowel in the early forms of Greek loan-words are the Plautine tĕchĭna, drăchŭma, Alcŭmĕna.

Later instances of Syncope, after the time when the Paenultima Law had come in, are calfacio, for calĕ-facio, originally calē-facio (§ 16), lamna, a bar of metal, from lammĭna (Plaut.), objūrgo, to scold, from objūrigo (Plaut.). Soldus (Hor. Sat. ii. 5. 65 metuentis reddere soldum), for sŏlĭdus, is the form that represented the ordinary pronunciation of every-day life, and caldus, we are told, was the form approved by the Emperor Augustus; though it was only in a few words that Syncope after a short syllable under the Paenultima Law was admitted into the literary language, e.g. valde, of which the full form vălide is still found in Plautus.

Syncope of the final syllable was a great feature of the Oscan and other Italic languages, but not of Latin. The Oscan word for 'Campanian,' for example, was Campans, while the Latin was Campānus; and Plautus mimics this Oscan pronunciation in his cruel sneer at the conquered Campanians (Trin. 545):

Campáns genus

Multó Surorum iam ántidit patiéntia,

'friend Campans is now far ahead of the Syrian race in endurance.' Final -rĭs syncopated in Adjs. like acer, Masc. of the stem ācri-, but this Syncope, like that of -rŏs in ager (Gk. ἀγρός), &c. (ch. iii. § 6), is pre-Latin, earlier than the separate existence of the Latin dialect of the Italic language. The -tĭs of words like nostrātis, 'our countryman,' of which the full forms only are known to Plautus, is usually said to have been reduced to -(t)s in nostrās, &c. by Syncope; but the new Nominative may rather be due to the analogy of a declension like amans Nom., amantis Gen., where -s in the Nom. case corresponds to -tis in the Gen. case. A Gen. Pl. like deum is sometimes called a 'contracted' form of deorum, but most mistakenly; for -um (Gk. -ωv) is the earlier suffix of the Gen. Pl. of the Second Declension, while -ōrum, formed on the analogy of -ārum of the First Declension, was an innovation introduced about the beginning of the literary period (see ch. iii. § 6). But a final -ĕ (-ĭ) was dropped in pronunciation, especially in words closely joined with the word that followed them, e.g. nec for neque,* ac* (i. e. *atc) for atque, neu for nēve, seu for sīve, quin for quīne (ch.ix. § 15), tot for *totĭ (cf. tŏtĭ-dem). Three Imperatives, which were much used in word-groups, lost their final -ĕ in the second century B.C., dīc, dūc, and fac. In Plautus the full forms are still used, when there is anything of a pause after the word; cf. Mil. 256 dice, monstra, praecipe, and especially Rud. 124:

tu, aíquid opus est, díce. Dic quod té rogo.

(On fer see ch. vi. § 14.)

The conditions under which Vowel-syncope was carried out differed at different periods. A vowel between n and m was not syncopated, because the consonant-group nm was difficult to pronounce, e. g. anima not anma, though in the Romance development of the Latin language we find that Syncope has been pushed a stage further, e. g. Old Fr. anme, alme, arme, Fr. âme, Span. alma, Ital. alma (in poetry). (So frigidus appears in all the Romance languages in a syncopated shape, e. g. Ital. freddo, Fr. froid, &c.). Analogy also may often prevent Syncope, or, after words have been syncopated, may restore them to their original form. Thus porgo, for example, was restored to porrigo by the analogy of the Perfect porrexi; and the analogy of other Adjectives in ‑ĭdus, where this termination was preceded by some uncombinable consonant, e. g. frigidus, may account for the existence of unsyncopated Adjectives like calidus; for the consonants in calidus, l and d, are of a kind that would easily combine. There are then two cases in which Latin vowels resisted Syncope: (1) when they stood between consonants which did not easily combine, (2) when in whole classes of words Syncope was prevented or effaced by the analogy of unsyncopated forms. With these exceptions it seems to have been the rule in Early Latin that ĕ, ĭ in the syllable after the accent always suffered syncope, unless they were long by ‘position.’ This ĕ, ĭ might be original ĕ, ĭ, or the reduced (posttonic) form of original ă (ŏ). The Early Latin accent fell, as we have seen, on the first syllable of each word, so that every ĕ, ĭ in a second syllable, not long by position, must have suffered Syncope.

The new law of accentuation, the Paenultima Law, brought with it the possibility of a new variety, namely, suppression of the syllable preceding the accent, Pretonic Syncope. It is often difficult to say whether a case of Syncope is pretonic or posttonic. In words like ārdere, ārdorem, for example, we say that the Syncope of ĭ of aridere, aridorem is due to the new accent on the penult, aridére, aridórem; but it might possibly be referred to the influence of the old accent on the first syllable, áridēre, áridōrem. Forms like artæna (Gk. ἀρύταινα), perstroma (Gk. περίστρωμα), both used by Lucilius, and both borrowed no doubt after the old Accent Law had ceased to operate, are clearer cases of Pretonic Syncope. And the influence of the following accent, rather than the mere addition of extra syllables, seems to be the real factor in the Syncope in the literary period of such words as frigidária (Lucil.) beside frígidus, caldárius beside cálidus, portórium beside pórtitor, postrídie beside pósterī, altrínsecus beside álteri. The unaccented ‑vĭ‑ of avidus, which resisted Syncope in the simple adjective‑form, succumbed to the influence of the following accent in the lengthened Derivative *avidere, audere, to dare, properly ‘to have a mind for’; cf. si audes, if you please (Plaut.), class. sōdes.

The analogy of these lengthened Derivatives, e. g. ardere, ardorem, caused or aided the Syncope of the simple Adjective, e. g. ardus (Lucil.). Similarly aet‑ for aevit‑ in aetas may have come into use first in the lengthened cases aetátis, aetáti, aetátem, or in Derivatives like aeternus. But in the literary period, as in the earlier, there was always the tendency to Syncope, and a word like aridus would, we may be sure, in the careless utterance of every‑day speech be pronounced ar'dus, though circumstances might operate in preventing this form from being accepted in literary Latin. Quintilian (i. 6. 19) tells us that Augustus stigmatized as a piece of affectation the use of calidus for caldus: non quia id non sit latinum, sed quia sit odiosum, et, ut ipse Graeco verbo significavit, περίεργον, and yet a Grammarian some centuries after (Appendix Probi 198. 3. K.) puts calda under the same condemnation as frigida, viridis.

Posttonic Syncope, under the new accent‑law, seems, during the Republic and Early Empire, to occur only when the accented vowel is long, e. g. bārca (our ‘barque’), from bārica, a word introduced at the time of Caesar’s naval displays in the Circus, although we find it in the period of the Early Literature in words of four or more syllables where three short syllables preceded the final, e. g. băĭlnĕum (Plaut. and Ter.), a spelling which did not yield for some time to later balneum. Opitumus was the form in use about the beginning of the literary period, but soon became optumus. These words, as we saw above (§ 12), had in the time of Plautus and Terence the accent on the first syllable, bálineum producing balneum, &c. Similarly opificina (Plaut.), accented òpificina, produced officina; pueritia, accented puéritia, produced the puertia of Horace (C. i. 36. 8). But forms like caldus from călidus, virdis from vĭridis, domnus for dŏminus, veclus for vetulus, are a feature of colloquial or Vulgar Latin, and were not as a rule established in the language till the later Empire; though valde, older vălide (Plaut.), and a few other words were current at a much earlier time.

The same wave of Syncope that reduced viridis, dominus, vetulus, &c., to disyllabic forms attacked u, i in hiatus (before a vowel). As early as the latter half of the first century A.D. tenuis varied between a disyllable and a trisyllable; cardus, for carduus, a thistle, mortus, for mortuus, &c., are the precursors of the Romance forms (Ital., Span. cardo; Ital. morto, Span. muerto, Fr. mort); while the similar reduction of i (e) led to that palatalization of consonants which has so transformed the whole appearance of the Romance languages, e. g. Ital. piazza, Span. plaza, Fr. place from Vulg. Lat. *platya, Lat. platĕa, &c. (see § 7).

The ă of ago remains unchanged in the simple Verb, where it has the accent, but in a compound like abigo it has by the classical period become ĭ. This change is not known in Greek (e. g. ἀπάγω like ἄγω), and is an effect of the Latin Stress-accent. In the period of the Early Literature we find ĕ instead of ĭ, e. g. abego (Plaut.), and before a labial ŭ, e. g. surrupui, surruptus (Plaut.) from sub and rapio; ĕ is the vowel still found in the classical age before r, e. g. peperi from pario, but memini from the root men-, or before a consonant-group, e. g. remex from remus and ago, princeps from primus and capio. Even diphthongs were changed through loss of stress, their first element being affected, ai became ī (through ei), au became ū (through eu), e. g. occido (earlier occeido) from caido (class. caedo), occlūdo from claudo. But not long vowels, e. g. invādo from vādo, irrēpo from rēpo, implōro from plōro. Unaccented ŏ became ŭ about the same time that ĕ passed in the unaccented syllable to ĭ; and ŭ, especially when an i followed in the next syllable, passed into ĭ or into that ü-sound which was written ĭ (§ 1); e. g. exsoles is the old Latin form of exules, from which comes exilium; quercubus became quercibus (cf. optumus and optimus, § 1). But ŏ, when not before a Labial, remains, e. g. in Compounds like invoco, advoco, and even before a Labial when a vowel precedes, e. g. filiolus [1]. In the final syllable it was invariably reduced [2], e. g. vīcus, older -os (cf. Gk. οἶκος), although after u or v the spelling with o was long retained to avoid the awkward collocation uu, e. g. vivos, divos, equos (written ecus in the Augustan Age, later equus, § 10).

[1] So ŏ remains after i, e. g. ebrietas, societas, parietem.

[2] So ĕ became ĭ, e. g. matris Gen. from *matrĕs, lĕgĭs 2 Sg. from *legĕs, legit 3 Sg. from *legĕt. But as a final vowel ĕ took the place of ĭ, e. g. mare for *marĭ, triste for *tristĭ (ch. iii. § 8), and perhaps of any short vowel. If a consonant is added, -ĕ becomes ĭ again, e. g. illĕ but illĭc (used for ille in the Dramatists). Hence sicine, hoccine, &c., for sī-c(ĕ), hoc-(cĕ) with the Interrogative -ne added, quippini for quippe with ni, and the like.

In the Imperial Age the fashion of preserving in Compounds the vowel of the Simple Verb came in, so that consacro, for example, was now written for consecro, the older spelling; and at the same time the Accent was, at least in the vulgar speech, shifted from the Preposition to the Verb; e. g. demórat is the Vulgar Latin form of demoratur, whence Ital. dimóra, Fr. demeure; from renégat, Ital. reniega (cf. Shakespeare's 'renége').

These changes of short vowels and diphthongs are proper to the syllable next the accent, the weakest syllable in every language with Stress-accentuation, that is to say to the second syllable of every word under the Early Accent Law (§ 12). A syllable with a secondary accent, like the paenultima of *párricaìda (so accented under this law), would not be liable to change, but often did in fact change its vowel after the analogy of kindred words, where the same vowel followed immediately on the accent, e. g. *óc-caido; whence O. Lat. paricidas (ch. iii. § 4). On the other hand, the analogy of the simple word with accented root-vowel would often save the vowel of the Compound from being changed, e. g. vades et subvades (XII Tab.), where the ă of vades is not weakened as it is in praevides (on an early inscr.), later praedes. And at any period in the history of the language the sense of the relation of a Compound to a simple word might lead to the restoration of a vowel to its accented quality. This 'Recomposition' was stimulated by the grammatical studies imported from Greece towards the close of the Republic, and prosecuted with zest for many centuries, so that in the period of the Early Literature, the change of unaccented vowels is more the rule than it is later, e. g. the weakening of the diphthong ae (ai) in O. Lat. forms like consīptum, obsīptum. The analogy of the Nominative preserved from change the vowel in the Oblique Cases of arborem, fulguris, &c., as on the other hand the analogy of the Oblique Cases has substituted ĕ for ĭ in the Nom. integer; and the analogy of the Compound Verb has changed the spelling of the little used Simple Verb in spicio, plico. Compounds, too, which were made for the occasion, or were rarely employed, like O. Lat. hosticapas, a conqueror of the enemy (ch. iii. § 4), urbi-capus (Plaut.), would escape the change which befel a word established in use, like prin-ceps, muni-ceps. But with these exceptions the change of the short vowels and the diphthongs of the second syllable is very regular in Latin; though the oldest inscription extant, Manios med fefaked [1] Numasioi, 'Manius me fecit Numerio,' belongs to an epoch when this law was not in operation, and when Latin was less removed from the state of the other Italic languages, which do not change the unaccented vowel (cf. Umbr. Propartio-, Lat. Propertius).

[1] The letter f is written ϝH (ch. i. § 4).

The usual course taken by the weakened vowel might be altered by other Phonetic Laws. It is, for example, a Phonetic Law of Latin that ĕ became ĭ before ng, e. g. tingo from *tengo (Gk. τέγγω) (ch. x. § 4). Hence we have infringo, confringo, and not *infrengo, *confrengo, &c., as compounds of frango. The same law of the influence of l on a preceding vowel that leaves velim, velle beside volo, volt (§ 8), gives us O. Lat. famelia as the earlier stage of familia, the Collective of famulus, O. Lat. *famolos. The Latin tendency to assimilate the vowel of neighbouring syllables (cf. momordī, from earlier memordi, ch. vi. § 10) was especially antagonistic to the Law of Weakening in Compounds like exemo, elego, neglego, the better spellings, and explains why we have surripio beside surrupui in MSS. of Plautus, and in class. Lat. incolumis (incolomis Plaut., &c.), monumentum (as well as monimentum), elementum, sepelio, coluber, segetis Gen., Seneca, tremebundus, alacris, aucupis Gen.

Examples of the older spelling of weakened vowels are:—in medial syllables (1) o for u: on early inscriptions, consoluerunt, consoleretur, consoltu, consol, consolibus, pocolom, conciliaboleis, in MSS. of Plautus exsolatum, incolomis, (2) u for i: on early inscriptions, Oinumama ('Unimamma,' an Amazon), testumonium, in MSS. of Plautus magnufice, sacruficem, carnufex. Manufestus, dissupo, victuma are the anteclassical, manifestus, dissipo, victima, the classical spellings, like optumus and optimus (§ 7), (3) e for i: on early inscriptions, meretod 'merito,' oppedeis 'oppidis,' in MSS. of Plautus abĕgit, of Lucretius accĕdo. We have in final syllables, e. g. O. Lat. cosentiont 'consentiunt,' dederont, nequīnont 'nequeunt,' opos, Venos, filios, Luciom, donom, Salutes Gen. Sg.

Other examples of the change in classical forms are:—(1) ĕ before a consonant-group: inlĕx from O. Lat. lacio, I allure, genetrix (beside genitor), obstetrix (cf. constituo), condemno (older condumno); expers from părs; perennis from ănnus; incestus from căstus; fefelli from făllo; miles (older miless, § 9) from stem milit-; remex Nom. from ago, (2) ĕ before r: aequipero, impero, pauper, all from paro; cineris Gen., (beside cinis Nom.), (3) ŭ, i before a Labial: incipio, decipio, and other Compounds of capio, surripui (also surpui). U remains in nuncupo, occupo, contubernium, which seem to be old forms preserved, while the o of vinolentus, somnolentus (cf. sanguinolentus) may be due to the analogy of vino lentus, &c., (4) ĭ in other short syllables: dimidius from medius, Jupiter (better spelt Juppiter) from pater, sistite (cf. Gk. ἵστατε), compitum 'ubi viae competunt,' dīmico from maco (cf. macto), in (earlier en) used enclitically (§ 11). U has become i in satira (and satura), inclitus (and inclutus), supercilium (cf. Gk. κύλα, the part under the eyes), but remains in tutudi, pecudem, contumax, &c. O has become u in venustus from Venus, older Venos, angustus, vetustus, and the like, alumnus (cf. Gk. τρεφόμενος), homullus from homŏn-lo-, &c., sum (earlier som), an enclitic (§ 11).

Examples in Greek loan-words are:—tessera (τέσσαρα), Agrigentum (Ἀκράγαντα Acc.), Hecuba, O. Lat. Hecoba (Ἑκάβη), machina (Dor. μαχανά), epistula and epistola (ἐπιστολή), Achivi (Ἀχαιοί), olivum (ἔλαιον), trutina (τρυτάνη), talentum (τάλαντον), phalerae (φάλαρα), balineum (Plaut.) and balneum (βαλανεῖον).

Long Vowels, as we have seen, did not suffer change. Delīro is the correct form, from līra, a furrow, not delēro (§ 10). The change of ē to ī in delīnio (beside delēnio), suspīcio (Noun) is found also in the case of accented ē when the following syllable has an i in hiatus, e. g. Plīnius from plenus, fīlius, lit. 'a suckling,' from fēlo. Anhēlus (older anhellus) has come from *an-ĕnslus, the a of hālo, from *ănslo, having been changed to e while its quantity was still short. Profestus is a compound of fēstus (cf. fēriae), not of fastus, fās (cf. nefastus).

Compounds with per, 'very,' are Separable Compounds [1] (like bene facio, ārē facio; cf. Lucr. facit are), and do not change the vowel, e. g. persalsus (beside insulsus), persapiens (beside insipiens), perfacilis (beside difficilis), so that Lucilius was right in his objection to the use of pertīsus by Scipio Africanus Minor:

Quó facetiór uideare et scíre plus quam céteri, 'Pértisum' hominem, nón pertaesum, dícis.

Ăv, ŏv become u, e. g. eluo from lavo, denuo for de novo.

[1] Cf. per pol saepe peccas, Plaut.

So sensitive was the Roman ear to the difference between a long and a short vowel (§ 1) that a long vowel was shortened through loss of stress only in special positions, namely in the final syllables of disyllabic words whose first syllable was short. In such a word as cắvē the Romans found themselves unable to maintain the length of the final ē, especially when the main stress of the voice rested on the initial syllable of the following word, e. g. in phrases like cave-fácias (§ 11). We ourselves, if we pronounce a phrase of this kind, feel that the short syllable că- exercises a shortening influence in this position upon the following long syllable -vē. The name usually now given to this law of shortening of Latin Vowels is the 'Law of the Breves Breviantes,' i. e. the law of the short (syllables) shortening (a following long syllable); and this law plays a great part in the prosody of the Republican Dramatists, whose verses reflect the pronunciation of every-day life, and so exhibit forms like căvĕ-fácias, vŏlŏ-scíre, as well as vŏlŭptátem, ĕgĕstátem, where the syllable scanned as a short syllable is long, not by nature, but by 'position.' It is this Law of Breves Breviantes which has effected the shortening of the final vowel of iambic words used in close connection with or subordination to other words, e. g. bĕnĕ, mălĕ (cf. bene-fácio, male-fácio, male-sánus), citō̆, modō̆, mihī̆, tibī̆, sibī̆. The finals of these words, though in poetical diction they often appear with the long quantity, were no doubt in ordinary speech by Cicero's time short or at least half-long vowels. Quintilian tells us that the -e of the salutation have (ave) was in his time pronounced long only by pedants, and the reduction of calē facio (properly written as two words) to calfacio points to an intervening stage when only calē facio was heard. From Auxiliaries like volo the shortening of the final -ō spread to other iambic Verbs, and finally, by analogy, to all Verbs in the first few centuries of the Empire, and the shortening of final -ā of Nouns (e. g. terra, originally *terrā), which was fully effected before the literary period (ch. iii. § 4), had taken the same course.

In the absence of stress, certain final consonants caused in the second century B.C. the shortening of preceding long vowels, viz. (1) -l, e. g. tribunăl Ovid, but bacchanāl Plaut., both from earlier -āle (§ 12), (2) -r, e. g. exemplăr Hor. from earlier exemplāre, class. mittăr, mittŏr, stridŏr, curĕr, but -ār, -ōr, -ēr in Plautus, (3) -t¹, e. g. class. mittăt, mittĕt, curăt, audĭt, dixĭt, but -āt, -ēt, -īt in Plautus. Before final -m too, which was in pronunciation dropped², giving the preceding vowel a nasal sound (§ 3), we know that the short quantity was used, e. g. rĕm (stem rē-).

¹ We find a similar difficulty in giving the o of 'note' as long a sound as the o of, let us say, 'node.'

² The dropping or weakening of final consonants in Latin was another result of the Stress-Accentuation. We find on plebeian inscriptions ama for amat, fecerun for fecerunt, &c. In the earlier poetry it is the rule, not the exception, that final s before an initial consonant does not lengthen a preceding short vowel by 'position,' but Cicero (Orator 48. 161) tells us that this pronunciation was in his time considered 'subrusticum.' It occurs only once in Catullus (cxvi. 8): tu dabi' supplicium.

Any long vowel or diphthong, even when accented, which preceded another vowel was reduced in quantity in Latin pronunciation, e. g. illīus became illĭus, praehendo became prĕhendo. So in the utterance of the sentence a final long vowel would be reduced before an initial vowel. This tendency is often reflected in poetry, e. g. quĭ amant Virg., especially in the older poetry, and no doubt contributed its share to the shortening of final -ā, -ō, which has just been mentioned. This explains why the Law of Breves Breviantes affected long vowels when final, but less frequently when preceding a final -s, -n, &c. (e. g. vĭdĕn).

One other case of the shortening of long vowels calls for mention, namely the shortening of the long monosyllable sī in sĭquidem, and (in the older poetry) of mē, tē, tū, in mē̆quidem, tē̆quidem, tū̆quidem, which has been compared to our shortening of 'sheep,' 'know,' &c., when a syllable is added, e. g. 'shep-herd,' 'know-ledge.' Probably the vowel did not, strictly speaking, become a short vowel, but changed its accent from a circumflex to an acute (§ 11).

§11

Accentuation

Accentuation

The rules of the Latin Accent are so simple that it was never found necessary to indicate by accent-marks the syllable on which the accent was to fall, as was done in Greek. In Latin the quantity of the penultimate syllable regulates accentuation, as the quantity of the final in Greek. When the penultimate syllable is long, the accent rests upon it; when short, on the antepenultimate, e. g. decōres, décǒres. This is also the natural English practice. There are however a few points of difficulty. Monosyllables with a long vowel, like flos, mos, res, had a 'Circumflex' Accent, the voice rising first and then falling slightly, and so had long final syllables of words whose last vowel has been dropt by Apocope or Syncope, e. g. illic, 'there,' from *illice, nostras, 'of our country,' from nostratis (Plaut.), audit for audivit, adduc for adduce, while monosyllables or apocopated finals with a naturally short vowel had the ordinary 'Acute' Accent, e. g. nix (nivis Gen.), pars, illinc, from *illim-ce. Again the Voc. and Gen. Sg. in -i of Nouns in -ius, -ium, were from the beginning of the Empire, or earlier, pronounced with the accent on the paenultima, even when the paenultima was short, e. g. Valéri, Vergíli, tugúri. This accentuation, introduced by Latin of Sentence writers on Grammar under the mistaken idea that these forms were contractions of an earlier Valérii, Vergílii, tugúrii, etc. (ch. iii. §6), was observed in the speech of the educated classes (cf. nostras). Further some of the usages of what is called 'Sentence-Accentuation' must be noticed, for the accent which a word would bear, if uttered separately, may be different from the accent assigned to it, when standing with other words in a sentence. The Greek Preposition πρός, for example, had, if mentioned by itself, an acute accent. But in the sentence its accent was obscured by the Noun which it governed, e. g. πρὸς πόλιν and this by the Greek system was expressed by replacing its acute with a grave accent. Similarly in Latin a Preposition, say súpra, or a Relative, say quális, were in sentences like supra moenia stat, talis est, qualis Cicero fuit, united with the following word into a word-group which took the ordinary accent of a single word, supra-moénia, qualis-Cícero. The Adverb supra and the Interrogative qualis were not subordinated, but retained their independent accent, súpra habitat, quális fuit Cicero? Other words which became subordinate or enclitic in the ordinary utterance of the sentence were:

(1) the various parts of the Substantive Verb, e. g. amatus-est (often written by the Romans amatust),

(2) the Personal and Possessive Pronouns, unless specially emphasized, e. g. tu-me, ad-me (cf. Gk. πρός με, πρός σε or our 'for him,' 'with him,' with the stress of the voice on 'for;' 'with,' unless the Pronoun is emphasized, as in 'for me, but not for thee'), meus-dóminus, mea-dómina (Ital. Madonna),

(3) the Demonstrative Pronouns, when unemphatic, e. g. ille-dóminus (pronounced with Syncope of -e, ill'-dominus); from this unemphatic use of the Demonstrative Pronoun has sprung the Romance Definite Article, e. g. Ital. 'il padre,' Span. 'el padre,'

(4) Conjunctions like et, sed,

(5) Auxiliary Verbs, such as volo in volo-scíre, cave in cave-fácias. Such Auxiliaries have in the languages derived from Latin been reduced to mere Tense-signs, e. g. Fr. 'j'aimerai' from Lat. amare-ha(be)o, 'j'ai fait' from Lat. ha(be)o-factum.

The subordination of many of these words was indicated by the Roman way of writing; the Preposition was often written-along with its Noun in one word, just as it was along with its Verb in a Compound Verb, incurrum, like incurro (cf. imprimis, obviam, admodum, denuo, for de novo, §15), while Auxiliaries like iubet, volo, were in certain cases similarly treated, e. g., quolibet, quamvis, quantumvis. And spellings like quomodo, quare, postridie (for < *posteri die), decemviri show that these Nouns, modus, res, dies, vir, were subordinated in Latin speech, just as in English 'thing,' 'kind,' 'part,' etc., are used without stress in phrases like 'something (nothing) of that kind,' 'some parts of England.' The Enclitics -que, -ve are examples of I.-Eur. Enclitics (cf. Gk. τε, *ϝε in ἠὲ rom *ἦ-ϝε). We saw above (§9) that the Latin pronunciation, like modern Italian, gave each syllable its due share of utterance, while the English slurs the unaccented syllables, and so has, for example, reduced 'fantasy' to 'fancy' and has given 'minute' (Lat. minutum) the sound of 'minit'. But this difference between Latin and English is not more than one of degree. For the Latin Accent was like ours a stress-accent; at all events it was an accent mainly of stress, though this stress was probably accompanied by a slightly higher tone than the tone of the unstressed syllables. And like all languages with stress-accent Latin had, though not to the same extent as English, the tendency to weaken the unstressed syllable, a tendency still seen in its modern representative, the Italian language (cf. Ital. 'gridare' from Lat. quiritare 'to shout,' 'shout for protection,' 'balsimo' from Lat. balsamum, albero from arborem). The Latin stress-accent left traces of itself at all periods of the language in the Syncope, the Weakening, and the Shortening of unaccented vowels, three processes which have so materially altered the appearance of the language that they must be treated in detail in separate paragraphs.

The rules of the Latin Accent are so simple that it was never found necessary to indicate by accent-marks the syllable on which the accent was to fall, as was done in Greek. In Latin the quantity of the penultimate syllable regulates accentuation, as the quantity of the final in Greek. When the penultimate syllable is long, the accent rests upon it; when short, on the antepenultimate, e. g. decōres, décǒres. This is also the natural English practice. There are however a few points of difficulty. Monosyllables with a long vowel, like flos, mos, res, had a 'Circumflex' Accent, the voice rising first and then falling slightly, and so had long final syllables of words whose last vowel has been dropt by Apocope or Syncope, e. g. illic, 'there,' from *illice, nostras, 'of our country,' from nostratis (Plaut.), audit for audivit, adduc for adduce, while monosyllables or apocopated finals with a naturally short vowel had the ordinary 'Acute' Accent, e. g. nix (nivis Gen.), pars, illinc, from *illim-ce. Again the Voc. and Gen. Sg. in -i of Nouns in -ius, -ium, were from the beginning of the Empire, or earlier, pronounced with the accent on the paenultima, even when the paenultima was short, e. g. Valéri, Vergíli, tugúri. This accentuation, introduced by Latin of Sentence writers on Grammar under the mistaken idea that these forms were contractions of an earlier Valérii, Vergílii, tugúrii, etc. (ch. iii. §6), was observed in the speech of the educated classes (cf. nostras). Further some of the usages of what is called 'Sentence-Accentuation' must be noticed, for the accent which a word would bear, if uttered separately, may be different from the accent assigned to it, when standing with other words in a sentence. The Greek Preposition πρός, for example, had, if mentioned by itself, an acute accent. But in the sentence its accent was obscured by the Noun which it governed, e. g. πρὸς πόλιν and this by the Greek system was expressed by replacing its acute with a grave accent. Similarly in Latin a Preposition, say súpra, or a Relative, say quális, were in sentences like supra moenia stat, talis est, qualis Cicero fuit, united with the following word into a word-group which took the ordinary accent of a single word, supra-moénia, qualis-Cícero. The Adverb supra and the Interrogative qualis were not subordinated, but retained their independent accent, súpra habitat, quális fuit Cicero? Other words which became subordinate or enclitic in the ordinary utterance of the sentence were:

(1) the various parts of the Substantive Verb, e. g. amatus-est (often written by the Romans amatust),

(2) the Personal and Possessive Pronouns, unless specially emphasized, e. g. tu-me, ad-me (cf. Gk. πρός με, πρός σε or our 'for him,' 'with him,' with the stress of the voice on 'for;' 'with,' unless the Pronoun is emphasized, as in 'for me, but not for thee'), meus-dóminus, mea-dómina (Ital. Madonna),

(3) the Demonstrative Pronouns, when unemphatic, e. g. ille-dóminus (pronounced with Syncope of -e, ill'-dominus); from this unemphatic use of the Demonstrative Pronoun has sprung the Romance Definite Article, e. g. Ital. 'il padre,' Span. 'el padre,'

(4) Conjunctions like et, sed,

(5) Auxiliary Verbs, such as volo in volo-scíre, cave in cave-fácias. Such Auxiliaries have in the languages derived from Latin been reduced to mere Tense-signs, e. g. Fr. 'j'aimerai' from Lat. amare-ha(be)o, 'j'ai fait' from Lat. ha(be)o-factum.

The subordination of many of these words was indicated by the Roman way of writing; the Preposition was often written-along with its Noun in one word, just as it was along with its Verb in a Compound Verb, incurrum, like incurro (cf. imprimis, obviam, admodum, denuo, for de novo, §15), while Auxiliaries like iubet, volo, were in certain cases similarly treated, e. g., quolibet, quamvis, quantumvis. And spellings like quomodo, quare, postridie (for < *posteri die), decemviri show that these Nouns, modus, res, dies, vir, were subordinated in Latin speech, just as in English 'thing,' 'kind,' 'part,' etc., are used without stress in phrases like 'something (nothing) of that kind,' 'some parts of England.' The Enclitics -que, -ve are examples of I.-Eur. Enclitics (cf. Gk. τε, *ϝε in ἠὲ rom *ἦ-ϝε). We saw above (§9) that the Latin pronunciation, like modern Italian, gave each syllable its due share of utterance, while the English slurs the unaccented syllables, and so has, for example, reduced 'fantasy' to 'fancy' and has given 'minute' (Lat. minutum) the sound of 'minit'. But this difference between Latin and English is not more than one of degree. For the Latin Accent was like ours a stress-accent; at all events it was an accent mainly of stress, though this stress was probably accompanied by a slightly higher tone than the tone of the unstressed syllables. And like all languages with stress-accent Latin had, though not to the same extent as English, the tendency to weaken the unstressed syllable, a tendency still seen in its modern representative, the Italian language (cf. Ital. 'gridare' from Lat. quiritare 'to shout,' 'shout for protection,' 'balsimo' from Lat. balsamum, albero from arborem). The Latin stress-accent left traces of itself at all periods of the language in the Syncope, the Weakening, and the Shortening of unaccented vowels, three processes which have so materially altered the appearance of the language that they must be treated in detail in separate paragraphs.

§12

Syncope of Unaccented Vowels

Syncope of Unaccented Vowels

For a century or two before the literary period the law of accentuation which was mentioned in the last paragraph, the 'Paenultima Law,' as it is called, did not prevail. Every word was at that time accented on the first syllable. Long words had, of course, as they had at all periods of the language, a secondary accent as well as the main accent [1], but in them, as well as in shorter words, the main accent rested on the first syllable.

[1] Cf. any English long word, say 'characteristical,' which has the main accent (´) on the antepenultimate, the secondary (`) on the first syllable.

The change to the Paenultima Law of Accentuation began in these long words with the substitution of the main accent for the secondary, the secondary for the main; e. g. témpestàtibus became tèmpestátibus, élemèntum became èleméntum (like our 'èleméntal'). The change to the Paenultima Law was not wholly completed at the beginning of the literary period, for the accent was still on the first syllable of words like fácilius, bálinĕum (later balneum), that is to say words of the scansion [short-short-short-short/long], in the time of Plautus. Instances of Syncope under the Early Accent Law are:

(1) of Prepositions in Compounds, e.g. anculus, an old word for 'servant' (whence ancilla), for *ambi-quolus (Gk. ἀμφί-πολος). By Syncope of this kind the old Preposition ambĭ was reduced to am- in Compounds ; similarly the old Preposition endŏ, indŏ, or indŭ was reduced to ind-, in-, and became confused with the Preposition in; e.g. induperator came to be pronounced as imperator, indugredi as ingredi;

(2) of the second syllable of the first member of a Compound, e.g. hospes for *hosti-pet-s, from hostis in its old sense of 'a stranger' ; princeps for *prīmi-ceps; forceps, the smith's tongs, from *formi-ceps from formus, hot, connected with Greek θερμός (ch. x. § 18); quindecim from quinque and decem; undecim, from ūnus and decem; vindemia, vintage, for vīnĭ-dēmia, from dēmo;

(3) of the first syllable of the Reduplicated Perfect of a Compound Verb, e. g. rettuli for re-tetuli (ch. vi. §10), repperi for re-peperi, reccĭdi or re-cecidi, where the double consonant preserves a trace of the Syncope.

When a short vowel following r was syncopated, the pronunciation of r without a vowel (as in the final syllable of Fr. 'sacre,' &c.) led to the form er. Thus *sacrŏ-dos, *sacrĭ-dos became sacerdōs; *sacrŏ-lom became *sacerlom, sacellum. Similarly -lĭ- appears in the syllable after the accent as ŭl (older ŏl) in facultas, &c. A syllable like -vĭ / vĕ- became after Syncope u in autŭmo for *ăvĭtumo (cf. Gk. οἴω for ὀϝ-ιω), auceps for *ăvĭ-ceps. But forms like lābrum for the older lăvābrum, dīnus for dīvīnus, audissem for audivissem, and the like, are not due to Syncope, but to the habit of dropping v in pronunciation between two vowels, especially similar vowels. Nor should we refer to Syncope such a change as dixti for dixisti, misse for misisse, debilitare for debilitatare, idolatria for idololatria, where, owing to the unpleasant effect of two neighbouring syllables having the same sound, one has been suppressed. *Saeclum *is not a syncopated form of saeculum, for saeclum is the older, while in saeculum a vowel has been inserted between the c and the l to aid pronunciation, like the parasitic ῐ of mina, a Greek loan-word, µvᾶ, with which we may compare Fr. 'canif,' a German loan-word, 'Knif,' or the parasitic u of Tecumessa (Τέκμησσα), for which Tecmessa was first used in Cicero's time [1]. On this ending -culus, -culum, and -clus, -clum, see ch. xi. § 9.

[1] Other examples of the Parasitic Vowel in the early forms of Greek loan-words are the Plautine tĕchĭna, drăchŭma, Alcŭmĕna.

Later instances of Syncope, after the time when the Paenultima Law had come in, are calfacio, for calĕ-facio, originally calē-facio (§ 16), lamna, a bar of metal, from lammĭna (Plaut.), objūrgo, to scold, from objūrigo (Plaut.). Soldus (Hor. Sat. ii. 5. 65 metuentis reddere soldum), for sŏlĭdus, is the form that represented the ordinary pronunciation of every-day life, and caldus, we are told, was the form approved by the Emperor Augustus; though it was only in a few words that Syncope after a short syllable under the Paenultima Law was admitted into the literary language, e.g. valde, of which the full form vălide is still found in Plautus.

Syncope of the final syllable was a great feature of the Oscan and other Italic languages, but not of Latin. The Oscan word for 'Campanian,' for example, was Campans, while the Latin was Campānus; and Plautus mimics this Oscan pronunciation in his cruel sneer at the conquered Campanians (Trin. 545):

Campáns genus

Multó Surorum iam ántidit patiéntia,

'friend Campans is now far ahead of the Syrian race in endurance.' Final -rĭs syncopated in Adjs. like acer, Masc. of the stem ācri-, but this Syncope, like that of -rŏs in ager (Gk. ἀγρός), &c. (ch. iii. § 6), is pre-Latin, earlier than the separate existence of the Latin dialect of the Italic language. The -tĭs of words like nostrātis, 'our countryman,' of which the full forms only are known to Plautus, is usually said to have been reduced to -(t)s in nostrās, &c. by Syncope; but the new Nominative may rather be due to the analogy of a declension like amans Nom., amantis Gen., where -s in the Nom. case corresponds to -tis in the Gen. case. A Gen. Pl. like deum is sometimes called a 'contracted' form of deorum, but most mistakenly; for -um (Gk. -ωv) is the earlier suffix of the Gen. Pl. of the Second Declension, while -ōrum, formed on the analogy of -ārum of the First Declension, was an innovation introduced about the beginning of the literary period (see ch. iii. § 6). But a final -ĕ (-ĭ) was dropped in pronunciation, especially in words closely joined with the word that followed them, e.g. nec for neque,* ac* (i. e. *atc) for atque, neu for nēve, seu for sīve, quin for quīne (ch.ix. § 15), tot for *totĭ (cf. tŏtĭ-dem). Three Imperatives, which were much used in word-groups, lost their final -ĕ in the second century B.C., dīc, dūc, and fac. In Plautus the full forms are still used, when there is anything of a pause after the word; cf. Mil. 256 dice, monstra, praecipe, and especially Rud. 124:

tu, aíquid opus est, díce. Dic quod té rogo.

(On fer see ch. vi. § 14.)

For a century or two before the literary period the law of accentuation which was mentioned in the last paragraph, the 'Paenultima Law,' as it is called, did not prevail. Every word was at that time accented on the first syllable. Long words had, of course, as they had at all periods of the language, a secondary accent as well as the main accent [1], but in them, as well as in shorter words, the main accent rested on the first syllable.

[1] Cf. any English long word, say 'characteristical,' which has the main accent (´) on the antepenultimate, the secondary (`) on the first syllable.

The change to the Paenultima Law of Accentuation began in these long words with the substitution of the main accent for the secondary, the secondary for the main; e. g. témpestàtibus became tèmpestátibus, élemèntum became èleméntum (like our 'èleméntal'). The change to the Paenultima Law was not wholly completed at the beginning of the literary period, for the accent was still on the first syllable of words like fácilius, bálinĕum (later balneum), that is to say words of the scansion [short-short-short-short/long], in the time of Plautus. Instances of Syncope under the Early Accent Law are:

(1) of Prepositions in Compounds, e.g. anculus, an old word for 'servant' (whence ancilla), for *ambi-quolus (Gk. ἀμφί-πολος). By Syncope of this kind the old Preposition ambĭ was reduced to am- in Compounds ; similarly the old Preposition endŏ, indŏ, or indŭ was reduced to ind-, in-, and became confused with the Preposition in; e.g. induperator came to be pronounced as imperator, indugredi as ingredi;

(2) of the second syllable of the first member of a Compound, e.g. hospes for *hosti-pet-s, from hostis in its old sense of 'a stranger' ; princeps for *prīmi-ceps; forceps, the smith's tongs, from *formi-ceps from formus, hot, connected with Greek θερμός (ch. x. § 18); quindecim from quinque and decem; undecim, from ūnus and decem; vindemia, vintage, for vīnĭ-dēmia, from dēmo;

(3) of the first syllable of the Reduplicated Perfect of a Compound Verb, e. g. rettuli for re-tetuli (ch. vi. §10), repperi for re-peperi, reccĭdi or re-cecidi, where the double consonant preserves a trace of the Syncope.

When a short vowel following r was syncopated, the pronunciation of r without a vowel (as in the final syllable of Fr. 'sacre,' &c.) led to the form er. Thus *sacrŏ-dos, *sacrĭ-dos became sacerdōs; *sacrŏ-lom became *sacerlom, sacellum. Similarly -lĭ- appears in the syllable after the accent as ŭl (older ŏl) in facultas, &c. A syllable like -vĭ / vĕ- became after Syncope u in autŭmo for *ăvĭtumo (cf. Gk. οἴω for ὀϝ-ιω), auceps for *ăvĭ-ceps. But forms like lābrum for the older lăvābrum, dīnus for dīvīnus, audissem for audivissem, and the like, are not due to Syncope, but to the habit of dropping v in pronunciation between two vowels, especially similar vowels. Nor should we refer to Syncope such a change as dixti for dixisti, misse for misisse, debilitare for debilitatare, idolatria for idololatria, where, owing to the unpleasant effect of two neighbouring syllables having the same sound, one has been suppressed. *Saeclum *is not a syncopated form of saeculum, for saeclum is the older, while in saeculum a vowel has been inserted between the c and the l to aid pronunciation, like the parasitic ῐ of mina, a Greek loan-word, µvᾶ, with which we may compare Fr. 'canif,' a German loan-word, 'Knif,' or the parasitic u of Tecumessa (Τέκμησσα), for which Tecmessa was first used in Cicero's time [1]. On this ending -culus, -culum, and -clus, -clum, see ch. xi. § 9.

[1] Other examples of the Parasitic Vowel in the early forms of Greek loan-words are the Plautine tĕchĭna, drăchŭma, Alcŭmĕna.

Later instances of Syncope, after the time when the Paenultima Law had come in, are calfacio, for calĕ-facio, originally calē-facio (§ 16), lamna, a bar of metal, from lammĭna (Plaut.), objūrgo, to scold, from objūrigo (Plaut.). Soldus (Hor. Sat. ii. 5. 65 metuentis reddere soldum), for sŏlĭdus, is the form that represented the ordinary pronunciation of every-day life, and caldus, we are told, was the form approved by the Emperor Augustus; though it was only in a few words that Syncope after a short syllable under the Paenultima Law was admitted into the literary language, e.g. valde, of which the full form vălide is still found in Plautus.

Syncope of the final syllable was a great feature of the Oscan and other Italic languages, but not of Latin. The Oscan word for 'Campanian,' for example, was Campans, while the Latin was Campānus; and Plautus mimics this Oscan pronunciation in his cruel sneer at the conquered Campanians (Trin. 545):

Campáns genus
Multó Surorum iam ántidit patiéntia,

'friend Campans is now far ahead of the Syrian race in endurance.' Final -rĭs syncopated in Adjs. like acer, Masc. of the stem ācri-, but this Syncope, like that of -rŏs in ager (Gk. ἀγρός), &c. (ch. iii. § 6), is pre-Latin, earlier than the separate existence of the Latin dialect of the Italic language. The -tĭs of words like nostrātis, 'our countryman,' of which the full forms only are known to Plautus, is usually said to have been reduced to -(t)s in nostrās, &c. by Syncope; but the new Nominative may rather be due to the analogy of a declension like amans Nom., amantis Gen., where -s in the Nom. case corresponds to -tis in the Gen. case. A Gen. Pl. like deum is sometimes called a 'contracted' form of deorum, but most mistakenly; for -um (Gk. -ωv) is the earlier suffix of the Gen. Pl. of the Second Declension, while -ōrum, formed on the analogy of -ārum of the First Declension, was an innovation introduced about the beginning of the literary period (see ch. iii. § 6). But a final -ĕ (-ĭ) was dropped in pronunciation, especially in words closely joined with the word that followed them, e.g. nec for neque,* ac* (i. e. *atc) for atque, neu for nēve, seu for sīve, quin for quīne (ch.ix. § 15), tot for *totĭ (cf. tŏtĭ-dem). Three Imperatives, which were much used in word-groups, lost their final -ĕ in the second century B.C., dīc, dūc, and fac. In Plautus the full forms are still used, when there is anything of a pause after the word; cf. Mil. 256 dice, monstra, praecipe, and especially Rud. 124:

tu, aíquid opus est, díce. Dic quod té rogo.

(On fer see ch. vi. § 14.)

§13

Details of Latin Syncope

Details of Latin Syncope

The conditions under which Vowel-syncope was carried out differed at different periods. A vowel between n and m was not syncopated, because the consonant-group nm was difficult to pronounce, e. g. anima not anma, though in the Romance development of the Latin language we find that Syncope has been pushed a stage further, e. g. Old Fr. anme, alme, arme, Fr. âme, Span. alma, Ital. alma (in poetry). (So frigidus appears in all the Romance languages in a syncopated shape, e. g. Ital. freddo, Fr. froid, &c.). Analogy also may often prevent Syncope, or, after words have been syncopated, may restore them to their original form. Thus porgo, for example, was restored to porrigo by the analogy of the Perfect porrexi; and the analogy of other Adjectives in ‑ĭdus, where this termination was preceded by some uncombinable consonant, e. g. frigidus, may account for the existence of unsyncopated Adjectives like calidus; for the consonants in calidus, l and d, are of a kind that would easily combine. There are then two cases in which Latin vowels resisted Syncope: (1) when they stood between consonants which did not easily combine, (2) when in whole classes of words Syncope was prevented or effaced by the analogy of unsyncopated forms. With these exceptions it seems to have been the rule in Early Latin that ĕ, ĭ in the syllable after the accent always suffered syncope, unless they were long by ‘position.’ This ĕ, ĭ might be original ĕ, ĭ, or the reduced (posttonic) form of original ă (ŏ). The Early Latin accent fell, as we have seen, on the first syllable of each word, so that every ĕ, ĭ in a second syllable, not long by position, must have suffered Syncope.

The new law of accentuation, the Paenultima Law, brought with it the possibility of a new variety, namely, suppression of the syllable preceding the accent, Pretonic Syncope. It is often difficult to say whether a case of Syncope is pretonic or posttonic. In words like ārdere, ārdorem, for example, we say that the Syncope of ĭ of aridere, aridorem is due to the new accent on the penult, aridére, aridórem; but it might possibly be referred to the influence of the old accent on the first syllable, áridēre, áridōrem. Forms like artæna (Gk. ἀρύταινα), perstroma (Gk. περίστρωμα), both used by Lucilius, and both borrowed no doubt after the old Accent Law had ceased to operate, are clearer cases of Pretonic Syncope. And the influence of the following accent, rather than the mere addition of extra syllables, seems to be the real factor in the Syncope in the literary period of such words as frigidária (Lucil.) beside frígidus, caldárius beside cálidus, portórium beside pórtitor, postrídie beside pósterī, altrínsecus beside álteri. The unaccented ‑vĭ‑ of avidus, which resisted Syncope in the simple adjective‑form, succumbed to the influence of the following accent in the lengthened Derivative *avidere, audere, to dare, properly ‘to have a mind for’; cf. si audes, if you please (Plaut.), class. sōdes.

The analogy of these lengthened Derivatives, e. g. ardere, ardorem, caused or aided the Syncope of the simple Adjective, e. g. ardus (Lucil.). Similarly aet‑ for aevit‑ in aetas may have come into use first in the lengthened cases aetátis, aetáti, aetátem, or in Derivatives like aeternus. But in the literary period, as in the earlier, there was always the tendency to Syncope, and a word like aridus would, we may be sure, in the careless utterance of every‑day speech be pronounced ar'dus, though circumstances might operate in preventing this form from being accepted in literary Latin. Quintilian (i. 6. 19) tells us that Augustus stigmatized as a piece of affectation the use of calidus for caldus: non quia id non sit latinum, sed quia sit odiosum, et, ut ipse Graeco verbo significavit, περίεργον, and yet a Grammarian some centuries after (Appendix Probi 198. 3. K.) puts calda under the same condemnation as frigida, viridis.

Posttonic Syncope, under the new accent‑law, seems, during the Republic and Early Empire, to occur only when the accented vowel is long, e. g. bārca (our ‘barque’), from bārica, a word introduced at the time of Caesar’s naval displays in the Circus, although we find it in the period of the Early Literature in words of four or more syllables where three short syllables preceded the final, e. g. băĭlnĕum (Plaut. and Ter.), a spelling which did not yield for some time to later balneum. Opitumus was the form in use about the beginning of the literary period, but soon became optumus. These words, as we saw above (§ 12), had in the time of Plautus and Terence the accent on the first syllable, bálineum producing balneum, &c. Similarly opificina (Plaut.), accented òpificina, produced officina; pueritia, accented puéritia, produced the puertia of Horace (C. i. 36. 8). But forms like caldus from călidus, virdis from vĭridis, domnus for dŏminus, veclus for vetulus, are a feature of colloquial or Vulgar Latin, and were not as a rule established in the language till the later Empire; though valde, older vălide (Plaut.), and a few other words were current at a much earlier time.

The same wave of Syncope that reduced viridis, dominus, vetulus, &c., to disyllabic forms attacked u, i in hiatus (before a vowel). As early as the latter half of the first century A.D. tenuis varied between a disyllable and a trisyllable; cardus, for carduus, a thistle, mortus, for mortuus, &c., are the precursors of the Romance forms (Ital., Span. cardo; Ital. morto, Span. muerto, Fr. mort); while the similar reduction of i (e) led to that palatalization of consonants which has so transformed the whole appearance of the Romance languages, e. g. Ital. piazza, Span. plaza, Fr. place from Vulg. Lat. *platya, Lat. platĕa, &c. (see § 7).

The conditions under which Vowel-syncope was carried out differed at different periods. A vowel between n and m was not syncopated, because the consonant-group nm was difficult to pronounce, e. g. anima not anma, though in the Romance development of the Latin language we find that Syncope has been pushed a stage further, e. g. Old Fr. anme, alme, arme, Fr. âme, Span. alma, Ital. alma (in poetry). (So frigidus appears in all the Romance languages in a syncopated shape, e. g. Ital. freddo, Fr. froid, &c.). Analogy also may often prevent Syncope, or, after words have been syncopated, may restore them to their original form. Thus porgo, for example, was restored to porrigo by the analogy of the Perfect porrexi; and the analogy of other Adjectives in ‑ĭdus, where this termination was preceded by some uncombinable consonant, e. g. frigidus, may account for the existence of unsyncopated Adjectives like calidus; for the consonants in calidus, l and d, are of a kind that would easily combine. There are then two cases in which Latin vowels resisted Syncope: (1) when they stood between consonants which did not easily combine, (2) when in whole classes of words Syncope was prevented or effaced by the analogy of unsyncopated forms. With these exceptions it seems to have been the rule in Early Latin that ĕ, ĭ in the syllable after the accent always suffered syncope, unless they were long by ‘position.’ This ĕ, ĭ might be original ĕ, ĭ, or the reduced (posttonic) form of original ă (ŏ). The Early Latin accent fell, as we have seen, on the first syllable of each word, so that every ĕ, ĭ in a second syllable, not long by position, must have suffered Syncope.

The new law of accentuation, the Paenultima Law, brought with it the possibility of a new variety, namely, suppression of the syllable preceding the accent, Pretonic Syncope. It is often difficult to say whether a case of Syncope is pretonic or posttonic. In words like ārdere, ārdorem, for example, we say that the Syncope of ĭ of aridere, aridorem is due to the new accent on the penult, aridére, aridórem; but it might possibly be referred to the influence of the old accent on the first syllable, áridēre, áridōrem. Forms like artæna (Gk. ἀρύταινα), perstroma (Gk. περίστρωμα), both used by Lucilius, and both borrowed no doubt after the old Accent Law had ceased to operate, are clearer cases of Pretonic Syncope. And the influence of the following accent, rather than the mere addition of extra syllables, seems to be the real factor in the Syncope in the literary period of such words as frigidária (Lucil.) beside frígidus, caldárius beside cálidus, portórium beside pórtitor, postrídie beside pósterī, altrínsecus beside álteri. The unaccented ‑vĭ‑ of avidus, which resisted Syncope in the simple adjective‑form, succumbed to the influence of the following accent in the lengthened Derivative *avidere, audere, to dare, properly ‘to have a mind for’; cf. si audes, if you please (Plaut.), class. sōdes.

The analogy of these lengthened Derivatives, e. g. ardere, ardorem, caused or aided the Syncope of the simple Adjective, e. g. ardus (Lucil.). Similarly aet‑ for aevit‑ in aetas may have come into use first in the lengthened cases aetátis, aetáti, aetátem, or in Derivatives like aeternus. But in the literary period, as in the earlier, there was always the tendency to Syncope, and a word like aridus would, we may be sure, in the careless utterance of every‑day speech be pronounced ar'dus, though circumstances might operate in preventing this form from being accepted in literary Latin. Quintilian (i. 6. 19) tells us that Augustus stigmatized as a piece of affectation the use of calidus for caldus: non quia id non sit latinum, sed quia sit odiosum, et, ut ipse Graeco verbo significavit, περίεργον, and yet a Grammarian some centuries after (Appendix Probi 198. 3. K.) puts calda under the same condemnation as frigida, viridis.

Posttonic Syncope, under the new accent‑law, seems, during the Republic and Early Empire, to occur only when the accented vowel is long, e. g. bārca (our ‘barque’), from bārica, a word introduced at the time of Caesar’s naval displays in the Circus, although we find it in the period of the Early Literature in words of four or more syllables where three short syllables preceded the final, e. g. băĭlnĕum (Plaut. and Ter.), a spelling which did not yield for some time to later balneum. Opitumus was the form in use about the beginning of the literary period, but soon became optumus. These words, as we saw above (§ 12), had in the time of Plautus and Terence the accent on the first syllable, bálineum producing balneum, &c. Similarly opificina (Plaut.), accented òpificina, produced officina; pueritia, accented puéritia, produced the puertia of Horace (C. i. 36. 8). But forms like caldus from călidus, virdis from vĭridis, domnus for dŏminus, veclus for vetulus, are a feature of colloquial or Vulgar Latin, and were not as a rule established in the language till the later Empire; though valde, older vălide (Plaut.), and a few other words were current at a much earlier time.

The same wave of Syncope that reduced viridis, dominus, vetulus, &c., to disyllabic forms attacked u, i in hiatus (before a vowel). As early as the latter half of the first century A.D. tenuis varied between a disyllable and a trisyllable; cardus, for carduus, a thistle, mortus, for mortuus, &c., are the precursors of the Romance forms (Ital., Span. cardo; Ital. morto, Span. muerto, Fr. mort); while the similar reduction of i (e) led to that palatalization of consonants which has so transformed the whole appearance of the Romance languages, e. g. Ital. piazza, Span. plaza, Fr. place from Vulg. Lat. *platya, Lat. platĕa, &c. (see § 7).

§14

Weakening of Unaccented Vowels

Weakening of Unaccented Vowels

The ă of ago remains unchanged in the simple Verb, where it has the accent, but in a compound like abigo it has by the classical period become ĭ. This change is not known in Greek (e. g. ἀπάγω like ἄγω), and is an effect of the Latin Stress-accent. In the period of the Early Literature we find ĕ instead of ĭ, e. g. abego (Plaut.), and before a labial ŭ, e. g. surrupui, surruptus (Plaut.) from sub and rapio; ĕ is the vowel still found in the classical age before r, e. g. peperi from pario, but memini from the root men-, or before a consonant-group, e. g. remex from remus and ago, princeps from primus and capio. Even diphthongs were changed through loss of stress, their first element being affected, ai became ī (through ei), au became ū (through eu), e. g. occido (earlier occeido) from caido (class. caedo), occlūdo from claudo. But not long vowels, e. g. invādo from vādo, irrēpo from rēpo, implōro from plōro. Unaccented ŏ became ŭ about the same time that ĕ passed in the unaccented syllable to ĭ; and ŭ, especially when an i followed in the next syllable, passed into ĭ or into that ü-sound which was written ĭ (§ 1); e. g. exsoles is the old Latin form of exules, from which comes exilium; quercubus became quercibus (cf. optumus and optimus, § 1). But ŏ, when not before a Labial, remains, e. g. in Compounds like invoco, advoco, and even before a Labial when a vowel precedes, e. g. filiolus [1]. In the final syllable it was invariably reduced [2], e. g. vīcus, older -os (cf. Gk. οἶκος), although after u or v the spelling with o was long retained to avoid the awkward collocation uu, e. g. vivos, divos, equos (written ecus in the Augustan Age, later equus, § 10).

[1] So ŏ remains after i, e. g. ebrietas, societas, parietem.

[2] So ĕ became ĭ, e. g. matris Gen. from *matrĕs, lĕgĭs 2 Sg. from *legĕs, legit 3 Sg. from *legĕt. But as a final vowel ĕ took the place of ĭ, e. g. mare for *marĭ, triste for *tristĭ (ch. iii. § 8), and perhaps of any short vowel. If a consonant is added, -ĕ becomes ĭ again, e. g. illĕ but illĭc (used for ille in the Dramatists). Hence sicine, hoccine, &c., for sī-c(ĕ), hoc-(cĕ) with the Interrogative -ne added, quippini for quippe with ni, and the like.

In the Imperial Age the fashion of preserving in Compounds the vowel of the Simple Verb came in, so that consacro, for example, was now written for consecro, the older spelling; and at the same time the Accent was, at least in the vulgar speech, shifted from the Preposition to the Verb; e. g. demórat is the Vulgar Latin form of demoratur, whence Ital. dimóra, Fr. demeure; from renégat, Ital. reniega (cf. Shakespeare's 'renége').

The ă of ago remains unchanged in the simple Verb, where it has the accent, but in a compound like abigo it has by the classical period become ĭ. This change is not known in Greek (e. g. ἀπάγω like ἄγω), and is an effect of the Latin Stress-accent. In the period of the Early Literature we find ĕ instead of ĭ, e. g. abego (Plaut.), and before a labial ŭ, e. g. surrupui, surruptus (Plaut.) from sub and rapio; ĕ is the vowel still found in the classical age before r, e. g. peperi from pario, but memini from the root men-, or before a consonant-group, e. g. remex from remus and ago, princeps from primus and capio. Even diphthongs were changed through loss of stress, their first element being affected, ai became ī (through ei), au became ū (through eu), e. g. occido (earlier occeido) from caido (class. caedo), occlūdo from claudo. But not long vowels, e. g. invādo from vādo, irrēpo from rēpo, implōro from plōro. Unaccented ŏ became ŭ about the same time that ĕ passed in the unaccented syllable to ĭ; and ŭ, especially when an i followed in the next syllable, passed into ĭ or into that ü-sound which was written ĭ (§ 1); e. g. exsoles is the old Latin form of exules, from which comes exilium; quercubus became quercibus (cf. optumus and optimus, § 1). But ŏ, when not before a Labial, remains, e. g. in Compounds like invoco, advoco, and even before a Labial when a vowel precedes, e. g. filiolus [1]. In the final syllable it was invariably reduced [2], e. g. vīcus, older -os (cf. Gk. οἶκος), although after u or v the spelling with o was long retained to avoid the awkward collocation uu, e. g. vivos, divos, equos (written ecus in the Augustan Age, later equus, § 10).

[1] So ŏ remains after i, e. g. ebrietas, societas, parietem.

[2] So ĕ became ĭ, e. g. matris Gen. from *matrĕs, lĕgĭs 2 Sg. from *legĕs, legit 3 Sg. from *legĕt. But as a final vowel ĕ took the place of ĭ, e. g. mare for *marĭ, triste for *tristĭ (ch. iii. § 8), and perhaps of any short vowel. If a consonant is added, -ĕ becomes ĭ again, e. g. illĕ but illĭc (used for ille in the Dramatists). Hence sicine, hoccine, &c., for sī-c(ĕ), hoc-(cĕ) with the Interrogative -ne added, quippini for quippe with ni, and the like.

In the Imperial Age the fashion of preserving in Compounds the vowel of the Simple Verb came in, so that consacro, for example, was now written for consecro, the older spelling; and at the same time the Accent was, at least in the vulgar speech, shifted from the Preposition to the Verb; e. g. demórat is the Vulgar Latin form of demoratur, whence Ital. dimóra, Fr. demeure; from renégat, Ital. reniega (cf. Shakespeare's 'renége').

§15

Details of Latin Vowel-weakening

Details of Latin Vowel-weakening

These changes of short vowels and diphthongs are proper to the syllable next the accent, the weakest syllable in every language with Stress-accentuation, that is to say to the second syllable of every word under the Early Accent Law (§ 12). A syllable with a secondary accent, like the paenultima of *párricaìda (so accented under this law), would not be liable to change, but often did in fact change its vowel after the analogy of kindred words, where the same vowel followed immediately on the accent, e. g. *óc-caido; whence O. Lat. paricidas (ch. iii. § 4). On the other hand, the analogy of the simple word with accented root-vowel would often save the vowel of the Compound from being changed, e. g. vades et subvades (XII Tab.), where the ă of vades is not weakened as it is in praevides (on an early inscr.), later praedes. And at any period in the history of the language the sense of the relation of a Compound to a simple word might lead to the restoration of a vowel to its accented quality. This 'Recomposition' was stimulated by the grammatical studies imported from Greece towards the close of the Republic, and prosecuted with zest for many centuries, so that in the period of the Early Literature, the change of unaccented vowels is more the rule than it is later, e. g. the weakening of the diphthong ae (ai) in O. Lat. forms like consīptum, obsīptum. The analogy of the Nominative preserved from change the vowel in the Oblique Cases of arborem, fulguris, &c., as on the other hand the analogy of the Oblique Cases has substituted ĕ for ĭ in the Nom. integer; and the analogy of the Compound Verb has changed the spelling of the little used Simple Verb in spicio, plico. Compounds, too, which were made for the occasion, or were rarely employed, like O. Lat. hosticapas, a conqueror of the enemy (ch. iii. § 4), urbi-capus (Plaut.), would escape the change which befel a word established in use, like prin-ceps, muni-ceps. But with these exceptions the change of the short vowels and the diphthongs of the second syllable is very regular in Latin; though the oldest inscription extant, Manios med fefaked [1] Numasioi, 'Manius me fecit Numerio,' belongs to an epoch when this law was not in operation, and when Latin was less removed from the state of the other Italic languages, which do not change the unaccented vowel (cf. Umbr. Propartio-, Lat. Propertius).

[1] The letter f is written ϝH (ch. i. § 4).

The usual course taken by the weakened vowel might be altered by other Phonetic Laws. It is, for example, a Phonetic Law of Latin that ĕ became ĭ before ng, e. g. tingo from *tengo (Gk. τέγγω) (ch. x. § 4). Hence we have infringo, confringo, and not *infrengo, *confrengo, &c., as compounds of frango. The same law of the influence of l on a preceding vowel that leaves velim, velle beside volo, volt (§ 8), gives us O. Lat. famelia as the earlier stage of familia, the Collective of famulus, O. Lat. *famolos. The Latin tendency to assimilate the vowel of neighbouring syllables (cf. momordī, from earlier memordi, ch. vi. § 10) was especially antagonistic to the Law of Weakening in Compounds like exemo, elego, neglego, the better spellings, and explains why we have surripio beside surrupui in MSS. of Plautus, and in class. Lat. incolumis (incolomis Plaut., &c.), monumentum (as well as monimentum), elementum, sepelio, coluber, segetis Gen., Seneca, tremebundus, alacris, aucupis Gen.

Examples of the older spelling of weakened vowels are:—in medial syllables (1) o for u: on early inscriptions, consoluerunt, consoleretur, consoltu, consol, consolibus, pocolom, conciliaboleis, in MSS. of Plautus exsolatum, incolomis, (2) u for i: on early inscriptions, Oinumama ('Unimamma,' an Amazon), testumonium, in MSS. of Plautus magnufice, sacruficem, carnufex. Manufestus, dissupo, victuma are the anteclassical, manifestus, dissipo, victima, the classical spellings, like optumus and optimus (§ 7), (3) e for i: on early inscriptions, meretod 'merito,' oppedeis 'oppidis,' in MSS. of Plautus abĕgit, of Lucretius accĕdo. We have in final syllables, e. g. O. Lat. cosentiont 'consentiunt,' dederont, nequīnont 'nequeunt,' opos, Venos, filios, Luciom, donom, Salutes Gen. Sg.

Other examples of the change in classical forms are:—(1) ĕ before a consonant-group: inlĕx from O. Lat. lacio, I allure, genetrix (beside genitor), obstetrix (cf. constituo), condemno (older condumno); expers from părs; perennis from ănnus; incestus from căstus; fefelli from făllo; miles (older miless, § 9) from stem milit-; remex Nom. from ago, (2) ĕ before r: aequipero, impero, pauper, all from paro; cineris Gen., (beside cinis Nom.), (3) ŭ, i before a Labial: incipio, decipio, and other Compounds of capio, surripui (also surpui). U remains in nuncupo, occupo, contubernium, which seem to be old forms preserved, while the o of vinolentus, somnolentus (cf. sanguinolentus) may be due to the analogy of vino lentus, &c., (4) ĭ in other short syllables: dimidius from medius, Jupiter (better spelt Juppiter) from pater, sistite (cf. Gk. ἵστατε), compitum 'ubi viae competunt,' dīmico from maco (cf. macto), in (earlier en) used enclitically (§ 11). U has become i in satira (and satura), inclitus (and inclutus), supercilium (cf. Gk. κύλα, the part under the eyes), but remains in tutudi, pecudem, contumax, &c. O has become u in venustus from Venus, older Venos, angustus, vetustus, and the like, alumnus (cf. Gk. τρεφόμενος), homullus from homŏn-lo-, &c., sum (earlier som), an enclitic (§ 11).

Examples in Greek loan-words are:—tessera (τέσσαρα), Agrigentum (Ἀκράγαντα Acc.), Hecuba, O. Lat. Hecoba (Ἑκάβη), machina (Dor. μαχανά), epistula and epistola (ἐπιστολή), Achivi (Ἀχαιοί), olivum (ἔλαιον), trutina (τρυτάνη), talentum (τάλαντον), phalerae (φάλαρα), balineum (Plaut.) and balneum (βαλανεῖον).

Long Vowels, as we have seen, did not suffer change. Delīro is the correct form, from līra, a furrow, not delēro (§ 10). The change of ē to ī in delīnio (beside delēnio), suspīcio (Noun) is found also in the case of accented ē when the following syllable has an i in hiatus, e. g. Plīnius from plenus, fīlius, lit. 'a suckling,' from fēlo. Anhēlus (older anhellus) has come from *an-ĕnslus, the a of hālo, from *ănslo, having been changed to e while its quantity was still short. Profestus is a compound of fēstus (cf. fēriae), not of fastus, fās (cf. nefastus).

Compounds with per, 'very,' are Separable Compounds [1] (like bene facio, ārē facio; cf. Lucr. facit are), and do not change the vowel, e. g. persalsus (beside insulsus), persapiens (beside insipiens), perfacilis (beside difficilis), so that Lucilius was right in his objection to the use of pertīsus by Scipio Africanus Minor:

Quó facetiór uideare et scíre plus quam céteri, 'Pértisum' hominem, nón pertaesum, dícis.

Ăv, ŏv become u, e. g. eluo from lavo, denuo for de novo.

[1] Cf. per pol saepe peccas, Plaut.

These changes of short vowels and diphthongs are proper to the syllable next the accent, the weakest syllable in every language with Stress-accentuation, that is to say to the second syllable of every word under the Early Accent Law (§ 12). A syllable with a secondary accent, like the paenultima of *párricaìda (so accented under this law), would not be liable to change, but often did in fact change its vowel after the analogy of kindred words, where the same vowel followed immediately on the accent, e. g. *óc-caido; whence O. Lat. paricidas (ch. iii. § 4). On the other hand, the analogy of the simple word with accented root-vowel would often save the vowel of the Compound from being changed, e. g. vades et subvades (XII Tab.), where the ă of vades is not weakened as it is in praevides (on an early inscr.), later praedes. And at any period in the history of the language the sense of the relation of a Compound to a simple word might lead to the restoration of a vowel to its accented quality. This 'Recomposition' was stimulated by the grammatical studies imported from Greece towards the close of the Republic, and prosecuted with zest for many centuries, so that in the period of the Early Literature, the change of unaccented vowels is more the rule than it is later, e. g. the weakening of the diphthong ae (ai) in O. Lat. forms like consīptum, obsīptum. The analogy of the Nominative preserved from change the vowel in the Oblique Cases of arborem, fulguris, &c., as on the other hand the analogy of the Oblique Cases has substituted ĕ for ĭ in the Nom. integer; and the analogy of the Compound Verb has changed the spelling of the little used Simple Verb in spicio, plico. Compounds, too, which were made for the occasion, or were rarely employed, like O. Lat. hosticapas, a conqueror of the enemy (ch. iii. § 4), urbi-capus (Plaut.), would escape the change which befel a word established in use, like prin-ceps, muni-ceps. But with these exceptions the change of the short vowels and the diphthongs of the second syllable is very regular in Latin; though the oldest inscription extant, Manios med fefaked [1] Numasioi, 'Manius me fecit Numerio,' belongs to an epoch when this law was not in operation, and when Latin was less removed from the state of the other Italic languages, which do not change the unaccented vowel (cf. Umbr. Propartio-, Lat. Propertius).

[1] The letter f is written ϝH (ch. i. § 4).

The usual course taken by the weakened vowel might be altered by other Phonetic Laws. It is, for example, a Phonetic Law of Latin that ĕ became ĭ before ng, e. g. tingo from *tengo (Gk. τέγγω) (ch. x. § 4). Hence we have infringo, confringo, and not *infrengo, *confrengo, &c., as compounds of frango. The same law of the influence of l on a preceding vowel that leaves velim, velle beside volo, volt (§ 8), gives us O. Lat. famelia as the earlier stage of familia, the Collective of famulus, O. Lat. *famolos. The Latin tendency to assimilate the vowel of neighbouring syllables (cf. momordī, from earlier memordi, ch. vi. § 10) was especially antagonistic to the Law of Weakening in Compounds like exemo, elego, neglego, the better spellings, and explains why we have surripio beside surrupui in MSS. of Plautus, and in class. Lat. incolumis (incolomis Plaut., &c.), monumentum (as well as monimentum), elementum, sepelio, coluber, segetis Gen., Seneca, tremebundus, alacris, aucupis Gen.

Examples of the older spelling of weakened vowels are:—in medial syllables (1) o for u: on early inscriptions, consoluerunt, consoleretur, consoltu, consol, consolibus, pocolom, conciliaboleis, in MSS. of Plautus exsolatum, incolomis, (2) u for i: on early inscriptions, Oinumama ('Unimamma,' an Amazon), testumonium, in MSS. of Plautus magnufice, sacruficem, carnufex. Manufestus, dissupo, victuma are the anteclassical, manifestus, dissipo, victima, the classical spellings, like optumus and optimus (§ 7), (3) e for i: on early inscriptions, meretod 'merito,' oppedeis 'oppidis,' in MSS. of Plautus abĕgit, of Lucretius accĕdo. We have in final syllables, e. g. O. Lat. cosentiont 'consentiunt,' dederont, nequīnont 'nequeunt,' opos, Venos, filios, Luciom, donom, Salutes Gen. Sg.

Other examples of the change in classical forms are:—(1) ĕ before a consonant-group: inlĕx from O. Lat. lacio, I allure, genetrix (beside genitor), obstetrix (cf. constituo), condemno (older condumno); expers from părs; perennis from ănnus; incestus from căstus; fefelli from făllo; miles (older miless, § 9) from stem milit-; remex Nom. from ago, (2) ĕ before r: aequipero, impero, pauper, all from paro; cineris Gen., (beside cinis Nom.), (3) ŭ, i before a Labial: incipio, decipio, and other Compounds of capio, surripui (also surpui). U remains in nuncupo, occupo, contubernium, which seem to be old forms preserved, while the o of vinolentus, somnolentus (cf. sanguinolentus) may be due to the analogy of vino lentus, &c., (4) ĭ in other short syllables: dimidius from medius, Jupiter (better spelt Juppiter) from pater, sistite (cf. Gk. ἵστατε), compitum 'ubi viae competunt,' dīmico from maco (cf. macto), in (earlier en) used enclitically (§ 11). U has become i in satira (and satura), inclitus (and inclutus), supercilium (cf. Gk. κύλα, the part under the eyes), but remains in tutudi, pecudem, contumax, &c. O has become u in venustus from Venus, older Venos, angustus, vetustus, and the like, alumnus (cf. Gk. τρεφόμενος), homullus from homŏn-lo-, &c., sum (earlier som), an enclitic (§ 11).

Examples in Greek loan-words are:—tessera (τέσσαρα), Agrigentum (Ἀκράγαντα Acc.), Hecuba, O. Lat. Hecoba (Ἑκάβη), machina (Dor. μαχανά), epistula and epistola (ἐπιστολή), Achivi (Ἀχαιοί), olivum (ἔλαιον), trutina (τρυτάνη), talentum (τάλαντον), phalerae (φάλαρα), balineum (Plaut.) and balneum (βαλανεῖον).

Long Vowels, as we have seen, did not suffer change. Delīro is the correct form, from līra, a furrow, not delēro (§ 10). The change of ē to ī in delīnio (beside delēnio), suspīcio (Noun) is found also in the case of accented ē when the following syllable has an i in hiatus, e. g. Plīnius from plenus, fīlius, lit. 'a suckling,' from fēlo. Anhēlus (older anhellus) has come from *an-ĕnslus, the a of hālo, from *ănslo, having been changed to e while its quantity was still short. Profestus is a compound of fēstus (cf. fēriae), not of fastus, fās (cf. nefastus).

Compounds with per, 'very,' are Separable Compounds [1] (like bene facio, ārē facio; cf. Lucr. facit are), and do not change the vowel, e. g. persalsus (beside insulsus), persapiens (beside insipiens), perfacilis (beside difficilis), so that Lucilius was right in his objection to the use of pertīsus by Scipio Africanus Minor:

Quó facetiór uideare et scíre plus quam céteri, 'Pértisum' hominem, nón pertaesum, dícis.

Ăv, ŏv become u, e. g. eluo from lavo, denuo for de novo.

[1] Cf. per pol saepe peccas, Plaut.

§16

Shortening of Unaccented Vowels

Shortening of Unaccented Vowels

So sensitive was the Roman ear to the difference between a long and a short vowel (§ 1) that a long vowel was shortened through loss of stress only in special positions, namely in the final syllables of disyllabic words whose first syllable was short. In such a word as cắvē the Romans found themselves unable to maintain the length of the final ē, especially when the main stress of the voice rested on the initial syllable of the following word, e. g. in phrases like cave-fácias (§ 11). We ourselves, if we pronounce a phrase of this kind, feel that the short syllable că- exercises a shortening influence in this position upon the following long syllable -vē. The name usually now given to this law of shortening of Latin Vowels is the 'Law of the Breves Breviantes,' i. e. the law of the short (syllables) shortening (a following long syllable); and this law plays a great part in the prosody of the Republican Dramatists, whose verses reflect the pronunciation of every-day life, and so exhibit forms like căvĕ-fácias, vŏlŏ-scíre, as well as vŏlŭptátem, ĕgĕstátem, where the syllable scanned as a short syllable is long, not by nature, but by 'position.' It is this Law of Breves Breviantes which has effected the shortening of the final vowel of iambic words used in close connection with or subordination to other words, e. g. bĕnĕ, mălĕ (cf. bene-fácio, male-fácio, male-sánus), citō̆, modō̆, mihī̆, tibī̆, sibī̆. The finals of these words, though in poetical diction they often appear with the long quantity, were no doubt in ordinary speech by Cicero's time short or at least half-long vowels. Quintilian tells us that the -e of the salutation have (ave) was in his time pronounced long only by pedants, and the reduction of calē facio (properly written as two words) to calfacio points to an intervening stage when only calē facio was heard. From Auxiliaries like volo the shortening of the final -ō spread to other iambic Verbs, and finally, by analogy, to all Verbs in the first few centuries of the Empire, and the shortening of final -ā of Nouns (e. g. terra, originally *terrā), which was fully effected before the literary period (ch. iii. § 4), had taken the same course.

In the absence of stress, certain final consonants caused in the second century B.C. the shortening of preceding long vowels, viz. (1) -l, e. g. tribunăl Ovid, but bacchanāl Plaut., both from earlier -āle (§ 12), (2) -r, e. g. exemplăr Hor. from earlier exemplāre, class. mittăr, mittŏr, stridŏr, curĕr, but -ār, -ōr, -ēr in Plautus, (3) -t¹, e. g. class. mittăt, mittĕt, curăt, audĭt, dixĭt, but -āt, -ēt, -īt in Plautus. Before final -m too, which was in pronunciation dropped², giving the preceding vowel a nasal sound (§ 3), we know that the short quantity was used, e. g. rĕm (stem rē-).

¹ We find a similar difficulty in giving the o of 'note' as long a sound as the o of, let us say, 'node.'

² The dropping or weakening of final consonants in Latin was another result of the Stress-Accentuation. We find on plebeian inscriptions ama for amat, fecerun for fecerunt, &c. In the earlier poetry it is the rule, not the exception, that final s before an initial consonant does not lengthen a preceding short vowel by 'position,' but Cicero (Orator 48. 161) tells us that this pronunciation was in his time considered 'subrusticum.' It occurs only once in Catullus (cxvi. 8): tu dabi' supplicium.

Any long vowel or diphthong, even when accented, which preceded another vowel was reduced in quantity in Latin pronunciation, e. g. illīus became illĭus, praehendo became prĕhendo. So in the utterance of the sentence a final long vowel would be reduced before an initial vowel. This tendency is often reflected in poetry, e. g. quĭ amant Virg., especially in the older poetry, and no doubt contributed its share to the shortening of final -ā, -ō, which has just been mentioned. This explains why the Law of Breves Breviantes affected long vowels when final, but less frequently when preceding a final -s, -n, &c. (e. g. vĭdĕn).

One other case of the shortening of long vowels calls for mention, namely the shortening of the long monosyllable sī in sĭquidem, and (in the older poetry) of mē, tē, tū, in mē̆quidem, tē̆quidem, tū̆quidem, which has been compared to our shortening of 'sheep,' 'know,' &c., when a syllable is added, e. g. 'shep-herd,' 'know-ledge.' Probably the vowel did not, strictly speaking, become a short vowel, but changed its accent from a circumflex to an acute (§ 11).

So sensitive was the Roman ear to the difference between a long and a short vowel (§ 1) that a long vowel was shortened through loss of stress only in special positions, namely in the final syllables of disyllabic words whose first syllable was short. In such a word as cắvē the Romans found themselves unable to maintain the length of the final ē, especially when the main stress of the voice rested on the initial syllable of the following word, e. g. in phrases like cave-fácias (§ 11). We ourselves, if we pronounce a phrase of this kind, feel that the short syllable că- exercises a shortening influence in this position upon the following long syllable -vē. The name usually now given to this law of shortening of Latin Vowels is the 'Law of the Breves Breviantes,' i. e. the law of the short (syllables) shortening (a following long syllable); and this law plays a great part in the prosody of the Republican Dramatists, whose verses reflect the pronunciation of every-day life, and so exhibit forms like căvĕ-fácias, vŏlŏ-scíre, as well as vŏlŭptátem, ĕgĕstátem, where the syllable scanned as a short syllable is long, not by nature, but by 'position.' It is this Law of Breves Breviantes which has effected the shortening of the final vowel of iambic words used in close connection with or subordination to other words, e. g. bĕnĕ, mălĕ (cf. bene-fácio, male-fácio, male-sánus), citō̆, modō̆, mihī̆, tibī̆, sibī̆. The finals of these words, though in poetical diction they often appear with the long quantity, were no doubt in ordinary speech by Cicero's time short or at least half-long vowels. Quintilian tells us that the -e of the salutation have (ave) was in his time pronounced long only by pedants, and the reduction of calē facio (properly written as two words) to calfacio points to an intervening stage when only calē facio was heard. From Auxiliaries like volo the shortening of the final -ō spread to other iambic Verbs, and finally, by analogy, to all Verbs in the first few centuries of the Empire, and the shortening of final -ā of Nouns (e. g. terra, originally *terrā), which was fully effected before the literary period (ch. iii. § 4), had taken the same course.

In the absence of stress, certain final consonants caused in the second century B.C. the shortening of preceding long vowels, viz. (1) -l, e. g. tribunăl Ovid, but bacchanāl Plaut., both from earlier -āle (§ 12), (2) -r, e. g. exemplăr Hor. from earlier exemplāre, class. mittăr, mittŏr, stridŏr, curĕr, but -ār, -ōr, -ēr in Plautus, (3) -t¹, e. g. class. mittăt, mittĕt, curăt, audĭt, dixĭt, but -āt, -ēt, -īt in Plautus. Before final -m too, which was in pronunciation dropped², giving the preceding vowel a nasal sound (§ 3), we know that the short quantity was used, e. g. rĕm (stem rē-).

¹ We find a similar difficulty in giving the o of 'note' as long a sound as the o of, let us say, 'node.'

² The dropping or weakening of final consonants in Latin was another result of the Stress-Accentuation. We find on plebeian inscriptions ama for amat, fecerun for fecerunt, &c. In the earlier poetry it is the rule, not the exception, that final s before an initial consonant does not lengthen a preceding short vowel by 'position,' but Cicero (Orator 48. 161) tells us that this pronunciation was in his time considered 'subrusticum.' It occurs only once in Catullus (cxvi. 8): tu dabi' supplicium.

Any long vowel or diphthong, even when accented, which preceded another vowel was reduced in quantity in Latin pronunciation, e. g. illīus became illĭus, praehendo became prĕhendo. So in the utterance of the sentence a final long vowel would be reduced before an initial vowel. This tendency is often reflected in poetry, e. g. quĭ amant Virg., especially in the older poetry, and no doubt contributed its share to the shortening of final -ā, -ō, which has just been mentioned. This explains why the Law of Breves Breviantes affected long vowels when final, but less frequently when preceding a final -s, -n, &c. (e. g. vĭdĕn).

One other case of the shortening of long vowels calls for mention, namely the shortening of the long monosyllable sī in sĭquidem, and (in the older poetry) of mē, tē, tū, in mē̆quidem, tē̆quidem, tū̆quidem, which has been compared to our shortening of 'sheep,' 'know,' &c., when a syllable is added, e. g. 'shep-herd,' 'know-ledge.' Probably the vowel did not, strictly speaking, become a short vowel, but changed its accent from a circumflex to an acute (§ 11).