A Latin Grammar for Schools and Colleges

A. Sound.

Alphabet.

16. In Cicero’s time, the sounds of the Latin language were denoted by twenty-one letters (DN. 2, 93).

Character	Name	pronounced
A	a	ah
B	be	bay
C	ce	kay
D	de	day
E	e	eh
F	ef	ef
G	ge	gay
H	ha	hah
I	i	ee
K	ka	kah
L	el	el
M	em	em
N	en	en
O	o	o
P	pe	pay
Q	qu	koo
R	er	air
S	es	ess
T	te	tay
V	u	oo
X	ix	eex

The names given above are those employed by Roman grammarians. The sound indicated by -ay is only approximate; the true sound is that of the French ê in fête; see 39. The names of the letters are indeclinable; for their gender, see 412.

17. Two other letters were also in use to represent Greek sounds in Greek words; these were always called by their Greek names, and were placed at the end of the alphabet; they are Y, named ü (42), and Z, named zēta (71).

18. Origin of the Alphabet. The Latin alphabet, which originally consisted of capitals only, was adapted from the alphabet of Chalcidian colonies in Italy.

19. Spelling. The signs for the Greek sounds denoted by φ and χ, and perhaps also that for θ, these three sounds being unknown in Latin, were used as numerals (2407). In words borrowed from the Greek the Romans at first represented θ by t, φ by p, and χ by c: as, tūs, incense, for θύος; Poenī, Punians, for Φοίνικες; calx, chalk, for χάλιξ. Occasionally also the Latin mute was doubled: as, struppus, strap, for στρόφος. Later, about the middle of the second century B.C., th, ph, and ch begin to be used: as, cothurnus, boot, for κόθορνος; amphora, jar, for ἀμφόρα; Achaea for Ἀχαιά. In some instances these aspirates were next introduced even into words purely Latin: as, chommodus, affable, for commodus, an affectation ridiculed by Catullus (Cat. 84) and disapproved by Quintilian (1, 5, 20). But pulcher, pretty, is the usual spelling for pulcer (formed by the suffix -cro- from the stem of the verb poliō, I polish). Even Cicero (O. 160) aspirated the c in this word as a concession to popular usage, as he did the t in Cethēgus, Karthāgō, and the p in triumphus, while he retained the unaspirated explosive in the proper names Orcīvius, name of a ‘gens,’ Matō, Otō, Caepiō, and in sepulcrum, tomb; corōna, crown; and lacrima, tear. In a similar manner Greek ρ was at first transcribed by r: as, rumpia, a kind of weapon, for ῥομφαία; but later by rh: as, rhētor, rhetorician, for ῥητωρ.

20. The letters C (first written <) and K were at an early period used promiscuously, and C stood for both unvoiced k and voiced g: as, VIRCO, virgō, virgin. Afterwards K dropped out of general use except in the abbreviations K. or Kal. for kalendae, first of the month, and K. for the proper name Kaesō (Quint. 1, 7, 10). About 300 B.C. the sign < or C was used for the unvoiced k alone, while a separate sign, which became G, was set apart for the voiced g. But C continued to be used for g in the abbreviations C for Gāius, Ↄ for Gāia, and Cn. for Gnaeus. Occasionally q is written for c, almost always before the vowels o and u: as, qum for cum, with; qolunt for colunt, they cultivate; peqūnia, money. But ordinarily q is found before unsyllabic (consonantal) u (v) only (22).

21. Before the introduction of Y and Z (17), u was used for the Greek υ: as, Burrus, later Pyrrhus (Cic. O. 160); and s, or, as a medial, ss, for ζ: as, sōna, belt, later zōna; massa, lump, for μᾶζα; malacissō, I soften, for μαλακίζω. By a blunder, y was occasionally introduced in words of Latin origin: as, lacryma, tear, for lacrima, which was wrongly supposed to be derived from Greek δάκρυ.

22. The characters I and V represent not only the two vowels i and u, but also their cognate semivowels (52) i̭ and ṷ (83), called commonly consonant i and u, but with less ambiguity unsyllabic i and u (82; 83). They are equivalent to the English y and w respectively.

23. In words like maior, simple i was commonly written for the sound of i̭i̭ (153, 2; 82; 83). But Cicero in such cases wrote ii: as, aiiō, I say, Maiia, Troiia (Quint. 1, 4, 11). In the same way Lucretius spelled Graiiugenārum, of Greek-born men, and EIIVS, of him, CVIIVS, whose, occur in inscriptions. Sometimes the same sound is represented by a taller letter, ‘i longa,’ especially in the imperial age: as, maIor, greater. There are also cases in which the two designations were confounded, a double i being written, and one or the other letter made taller: as, eiIvs or eIivs, of him.

24. The tall i, I longa, was used not only to represent unsyllabic i (22), but, beginning with Sulla’s time, also for long vowel i (29, 2, b): as, sIgna, signs; qvInqve, five. It also represents sometimes double i: as, vIs for viīs, in the roads. At the beginning of words it occurs without reference to quantity for both short and long i, and, by mistake, I is elsewhere found for short i.

25. The emperor Claudius (A.D. 41-54) introduced a separate sign for unsyllabic u (22), restricting the sign v to the vowel u (Quint. 1, 7, 26; Ta. 11, 14); but it did not become current.

26. In schoolbooks and most texts of the authors, the vowel u is printed U, u, and the consonant V, v. A character, J, j, was introduced in the 17th century, to indicate the consonant i. But this character is no longer usual in editions of the authors or in schoolbooks.

27. The distinction between u and v is not always made very consistently: q has regularly, and g and s have sometimes, an aftersound of w, best represented by v; but the usual practice is to write u, as in the following disyllables: quōrum, of whom; anguis, snake; suāvis, sweet. qu is always counted as a single sound (177). See also 2504.

28. For the intermediate sound (103) between i and u, as in the first syllable of lubet, libet, it pleases, and in the second syllable of optimus, optumus, best (Quint. 1, 4, 8; 7, 21), the emperor Claudius invented a separate character. It failed of acceptance, as did also the sign which he attempted to introduce for ps.

29. The same characters were ordinarily used to denote both long and short vowels. But at different periods long vowels were sometimes indicated in inscriptions thus:

(1.) Long a, e, or u was sometimes doubled: as, AARA, altar; PAASTORES, shepherds; LEEGE, by law; IVVS, right. This doubling, which was never frequent, seems to have been introduced into Latin from the Oscan by the poet Accius. It occurs most frequently in inscriptions about the year 150 B.C., but sporadically much later: as, CONVENTVVS, of the assembly; ARBITRATVV, by the decree; and in other stems in -u- (593).

(2.) Long i was often denoted (a.) By the spelling ei (after the pronunciation of this diphthong had been changed to ī, 98): as, DAREI, be given; REDIEIT, hath come back; INTERIEISTI, hast died. Some Roman grammarians prescribed this spelling for every long i; others tried to regulate the use of ei for ī by special rules. At the end of the republic, the spelling EI had given way to uniform I. (b.) Since the time of Sulla, by a taller letter (‘i longa’): as, fIxa, fastened (23, 24).

(3.) A mark called an apex ([three apices]) was often put over a long vowel: as, FE͆CIT, made; HORTE͆NSIVS; DVV͆MVIRATVS, duumvirate. The apex was written ´ in the imperial age; the form -, which occurs in an inscription, was adopted by the grammarians, and is still in use to mark the long vowels. It may be mentioned that inscriptions which employ the apex are by no means consistent in its use, and that late inscriptions have it over short and long vowels, apparently for decorative purposes. Quintilian 1, 7, 2 prescribes it only for cases which otherwise might be ambiguous: as, MÁLVS (mālus), mast, to distinguish it from MALVS (malus), bad.

30. In schoolbooks, a long vowel is indicated by a horizontal line over it: as, āra, altar; mēnsis, month; ōrdō, series. A short vowel is sometimes indicated by a curved mark: as, pĕr, through; dŭx, leader; but this mark is unnecessary if long vowels are systematically marked. Usually the quantity of the vowels in each word is definitely fixed; but in a few cases the same vowel may be now short, now long, as in English the ee of been is pronounced long by some (bean), short by others (bin). Thus (2446) mihi, ibi were sometimes pyrrhics (⏖, 2522), sometimes iambi (⏑-, 2521). See for other cases 134, 2443, 2452, 2453. Such vowels of variable quantity are termed common and marked ⏓ or ⏒: as mihī̆, to me (2514).

Pronunciation.

31. The pronunciation of Latin sounds may be approximately determined: (a) from the description of the native grammarians and incidental allusions in other Latin authors; (b) from variations in spelling; (c) from the Greek transliteration of Latin words; (d) from the Latin transliteration of foreign words; (e) from the development of the sounds in languages derived from the Latin.

Vowels.

32. Vowels are sounds which are produced by the vibrations of the vocal chords (this may be easily felt by placing a finger on the throat at the Adam’s apple) and without any audible friction or any obstruction anywhere in the passage above the vocal chords. The difference in the sound of the vowels is due to the different shape which the position of the tongue and the lips gives in each case to the cavity of the mouth. During the pronunciation of pure vowels no air escapes through the nose.

33. The simple vowels, a, e, i, o, u (y), are either long or short. The sound of a long vowel is considered to be twice the length of that of a short.

34. That a long vowel is equal to two shorts is a rule of metrical theory (see 2515). In actual pronunciation, there were undoubtedly various degrees of length, as in English: e.g., sea, seize (long), cease (half-long).

Quantity of Vowels.

The quantity of vowels must in general be learned by observation; but some convenient helps for the memory may be found in 2429; and the quantity of many vowels may be ascertained by the general principles given in 35 and 36. Except in the case of Hidden Quantity (2459), the quantity of vowels is in general ascertained from verse. But some information may also be gleaned from such rhetorical prose as exhibits well defined habits in the rhythmical endings selected for sentences (clausulae, Cic. O. 191-226).

(A.) SHORT VOWELS.

35. A vowel is short:

(1.) Before another vowel or h (124): as, eōs, ēvehō; compare taceō with tacēre. For exceptions in classical Latin, see 127; for exceptions in early Latin see 126.

(2.) Before nt and nd (128) if not the result of contraction: as, calendae, centum; compare amant, amandus, with amāre.

(3.) Before final t and m, and, in words of more than one syllable, before final r and l (132): compare amat, amem, with amās and amēs.

(B.) LONG VOWELS.

36. All vowels are long which are:

(1) Weakened from a diphthong (96-101; 108), or which are the result of contraction (118): as, concīdō from caedō; cōgō from co-agō.

(2) Lengthened by compensation (121): as, quīnī for *quincnī.

(3) Before nf, ns, often before nc followed by a consonant, and, in some cases, before gn (122).

Pronunciation of Vowels.

37. The following English sounds come nearest to the Latin pronunciation of the vowels:

38. Long vowels. ā had the sound of a in father; ē that of a in fate (but see 39); ī that of i in machine; ō that of o in tone; ū that of u in rule.

39. It must be noted, however, that all English long vowels, save a as in father, are more or less diphthongal, that is, they become gradually closer (46); a in fate ends in a vanishing sound of ee (not heard in the ê of French fête), and o in no ends in the sound of oo. Similarly the long e sound in he becomes closer and ends in a sound similar to the y in year. In Latin all long vowels had one sustained sound.

40. Short vowels. a sounded approximately like the English a in the first syllable of aha; e, i, o, and u sounded like e in step, i in pit, o in obey, and u in pull respectively.

41. Latin short a did not differ, except in quantity, from long ā; it never had the ‘flat’ sound of English a in pat. In the case of the other vowels, i, e, o, and u, the long vowels were closer (46) than the short ones. This is the same difference which the English shows in keen (long and close) and kin (short and open); pool (long and close) and pull (short and open). For this reason, open i is sometimes represented by e in inscriptions: as, ANEMA for anima, soul; and vea was the rustic pronunciation for via, road (Varro, R. R. 1, 2, 14).

42. Y, which was a sound borrowed from the Greek (17), sounded like German ü. The sound, which is missing in English, is formed with the tongue in position for i (in kin) and the lips rounded as for oo (in moon).

Classification of Vowels.

43. Vowels are divided according to the position of the tongue. Latin i and e are called front vowels, because the front part of the tongue is elevated. This elevation is greater for i than for e. Latin o and u are called back vowels, because they require an elevation of the rear part of the tongue. This elevation is greater for u than for o. Latin a holds an intermediate position, no part of the tongue being raised, while the front part is depressed.

44. In the formation of i and e, the tongue approaches the hard palate; hence these two vowels are also called palatal vowels. Similarly, o and u are called velar or guttural vowels, because in their formation the tongue approaches the soft palate (vēlum palātī).

45. o and u require a rounding of the lips (labia); hence they are called labial vowels. The same is true for y.

46. Comparing the vowels in English keen and kin, it will be noted that the passage between the tongue and the hard palate is narrower in the former than in the latter case. The ee in keen is therefore said to be a narrow or close vowel, while the i in kin is wide or open. See 41.

Diphthongs.

47. Two unlike (43-46) vowels pronounced under one stress and as one syllable form a Diphthong. All diphthongs are long.

In all diphthongs the transition from one vowel to the other is gradual. A diphthong is, therefore, not formed simply by pronouncing two vowels in succession, but the vocal organs pass through all the intermediate positions and consequently the sound is constantly changing.

48. In their origin diphthongs are of two kinds: (a.) primitive diphthongs: as in foedus, treaty; aurum, gold; or (b.) secondary diphthongs, the result of vowels meeting in formation, composition, or inflection: see 120.

49. The diphthongs which occur in classical Latin are au, ae, oe, and the rare ui and eu.

au sounded like ou in house. ae had the sound of short Latin a rapidly combined with the sound of e in English men. But it is the common practice now to give to ae the sound of ay or ai in ay, aisle, although the difference between Latin ae and the earlier ai from which it descended is thus obliterated. oe had the sound of short Latin o rapidly followed by the sound of e in English men. But it is now customary not to distinguish between Latin oe and oi, and to give to both the sound of oi in boil. ui is pronounced by combining Latin short u and i (40, 41) with the stress on the i like French oui; eu by combining Latin short e and u with stress on the u.

50. Besides these, the following diphthongs occur in the older inscriptions: ai pronounced as ai in aisle; ei as ei in eight; oi as oi in boil; and ou which sounded very much like the final o in no, go, which is really a diphthong (see 39).

Consonants.

51. Consonants are formed by stopping the breath somewhere in the cavity of the mouth or by squeezing it through a narrow channel or aperture.

52. Semivowels. There is no sharp line of demarcation between consonants and vowels. Some vowels in unsyllabic function (82, 83) notably i (i̭) and u (ṷ) (corresponding to English y and w), though usually classed as consonants, are so closely related to the vowels that they are termed semivowels (2504). To these may be added also the liquids l and r. Contact of the semivowels i and u with their corresponding vowels i and u is avoided in classical times. See for -vu- 107, c; for -quu- 157; and for -i̭i- 104, c (on obi̭iciō); 458 (Bōī for *Bōi̭ī). See 153, 3.

Pronunciation of Consonants.

53. Most of the consonants are pronounced as in English. The following points must be noticed:

54. b before a surd, as s or t, has the sound of p. The spelling b is here simply etymological: as, abs, pronounced aps (the b retained in spelling because of ab); urbs, pronounced urps (the b retained because of the oblique cases urbis, urbī, etc.); obterō, pronounced opterō (Quint. 1, 7, 7), where the spelling of the preposition ob was kept (164).

55. c has always the sound of English k.

56. d before the surd s is pronounced t; the spelling d is preserved for etymological reasons only: as, adsum, pronounced atsum.

57. g always has the sound of English g in go, never that of g in gentle. gu, when it makes one syllable with the following vowel, is pronounced like English gw: as, sanguine like sanguine.

58. h has a weak sound as h in British English (Southern), and by some was not counted as a consonant. Consequently the same uncertainty existed as to initial h. The omission of initial h is recognized in classical Latin for ānser (originally *hānser). Elsewhere the omission of initial h in spelling, as ostia for hostia, is rare until the third century A.D.

Very rarely h is written between two vowels to denote that each should be pronounced separately (like our diaeresis in coëxtensive): as, ahēneus, bronze, with aē separate (116 a); but aes, bronze, with diphthongal ae.

59. Unsyllabic (22) or consonant i has the sound of English y in year.

60. There were two varieties of l. One was like the English l, guttural in character, because in its pronunciation not only the blade (front part) of the tongue touched the gums, but in addition to this the rear part of the tongue was elevated toward the soft palate. The other l was purely dental, and formed without such back elevation. This second variety appeared in the combination ll, or whenever l was followed by the front vowels (43) e or i, or when it was final. Elsewhere l was guttural.

61. From the earliest times final m in unaccented syllables had a faint sound or was even inaudible (Quint. 9, 4, 39). Consequently it is often omitted in writing in the older inscriptions both before an initial vowel or consonant: as, POCOLO for pōcolom; OINO for oinom (ūnum), and the grammarian Verrius Flaccus proposed to write only half an M for final m before a vowel. In prosody, therefore, final m did not prevent elision (2493). The same is seen in prose in cases like animadvertō, I pay heed to, from anim^um advertō, I turn my mind toward (395); vēnīre, to be sold for vēn^um īre, to go to sale (1165). But in monosyllables where m closes the accented syllable, it did not vanish (2494, 2495), and this difference in the treatment of final m is reflected in the Romance languages.

62. n stands for two sounds. It represents the dental nasal, as n in English now. But before the gutturals k, c, g, q, and the compound x (= cs), it represents the guttural nasal which is written ng in English sing, wrong. This second n is sometimes called n adulterīnum or ‘spurious n,’ thus: nc (in avunculus) as in uncle; ng (in angulus) as in angle; ngu (in sanguine) as in sanguine; nqu (in inquit) as inkw in inkwiper; nx (in pīnxit) as in lynx.

63. Dental n before s had a reduced sound, and is therefore sometimes omitted in writing: as, CESOR for cēnsor; COSOL for cōnsul, in older inscriptions; and fōrmōsus by the side of fōrmōnsus; vīcēsimus by the side of vīcēnsimus, Cicero omitted the n in the adjective suffix -ēnsis: as, forēsia, of the forum; hortēsia, garden plants.

64. q, in classical Latin, appears only in the combination qu, sounded like English qu or kw (27). r was trilled.

65. s, in classical Latin was always unvoiced (surd, 75) like English s in so, sin, never voiced (sonant, 75) as English s in ease. su, when it makes one syllable with the following vowel, is like sw in sweet (27).

66. In old Latin, final s after a short vowel and before a consonant seems to have been reduced in sound or to have disappeared altogether. In the older inscriptions it is often omitted in the ending of the nominative singular -us, and in the pre-Ciceronian poets final s often does not make position (2468). But such omission was considered vulgar in Cicero’s time (Cic. O. 161; Quint. 9, 4, 38).

67. In the archaic period Latin s stood also for the voiced sibilant (English s in ease, z in zeal), as in ASA, altar (154).

68. t is always sounded as in time, never as in nation. The pronunciation of ci and ti with the c and t as sibilants (as in English cinder, nation) is very late.

69. v is like the English w.

70. x is a compound consonant, standing for cs, and so sounded, never as English gs or gz.

71. z, being a Greek sound, should have retained its Greek pronunciation. This differed in the different dialects; in the Attic of the fourth century B.C. it was approximately that of English z in zeal, while its earlier value was zd. The Romans had great difficulty in pronouncing this sound (Quint. 12, 10, 27 f.), but the grammarian Velius Longus expressly states that it should not be pronounced as a compound sound (zd).

72. About 100 B.C. the combinations ch, ph, and th were introduced in Greek words to represent χ, φ, and θ; as Philippus, for the older PILIPVS. Somewhat later these combinations were in general use in some Latin words (19). ch is thought to have been pronounced like kh in blockhead, ph as in uphill, and th as in hothouse. But in practice ch is usually sounded as in the German machen or ich, ph as in graphic, and th as in pathos.

Classification of Consonants.

73. Explosives. Consonants which are formed by stopping the breath in the oral cavity and then suddenly removing the obstruction are called explosives. They cannot be prolonged in sound. They are: c, k, q, g; t, d; p, b. These are often called mutes.

74. Continuants. Consonants which may be prolonged in sound are called continuants. They are: unsyllabic (83) i (59) and u (66); l (60), r; l, s, f; n (62), m.

75. Voiced and Unvoiced. If during the emission of breath the vocal chords vibrate (32), the consonant is said to be voiced or sonant: g; d; b; n (62), m; l (60), r; unsyllabic (83) i (59) and u (69); otherwise it is said to be unvoiced or surd: c, k, q; t; p; h, s, f.

76. Nasals. In the majority of consonants, the breath escapes through the cavity of the mouth, and the cavity of the nose is closed in the rear by means of the raised soft palate. Those consonants in which the breath escapes through the nose, while the oral cavity is closed, are called nasals: as, n, m, n adulterīnum (see 62).

77. Classification according to place of formation. Consonants are further divided according to the place where the breath is stopped or squeezed. (1.) If the breath is stopped by the lips, as in p, b, m, or squeezed through the lips, as in v (English w), we speak of labials. (2.) If the breath is forced through an opening between the upper teeth and the lower lip, as in f, we speak of a labiodental. (3.) Sounds which are produced by the point of the tongue touching the upper gums and teeth, as t, d, n, r, or by the formation of a narrow median channel in the same place, like s, or of a lateral channel, like l (60), are called dentals. (4.) Palatals are formed by an elevation of the front part of the tongue against the forward section of the palate, like i consonant (English y). (5.) If the back of the tongue touches or approaches the rear part of the palate as in k, q, c, g, n adulterīnum (English ng in sing), and l (60), we speak of gutturals (velars); see 44.

78. Spirants. Sounds which are produced by friction of the breath are called spirants: as, s, f, and h.

79. Sibilants. On account of its hissing sound, s is called a sibilant. English s, z, th are sibilants.

80. Doubling of Consonants. In English, double consonants as the tt, nn, pp, mm in motto, Anna, tapping, grammar, are sounded exactly like the corresponding single consonants in cot, pan, tap, ram. In Latin, on the other hand, double consonants (geminātae) were pronounced as they are in modern Italian. In the case of explosives (73), as in mitto, after the tongue had come in contact with the roof of the mouth (= first t) a short pause ensued before the explosion took place (= second t). In the case of continuants (74), as in summus, Apollo, the mm or ll was sounded appreciably longer than a single m or l, and at the beginning of the second half of the long continuant there was a slight increase of force.

81. Consonants were not doubled in writing till after 200 B.C.: as, FVISE for fuisse, to have been, and for more than a century afterward the usage is variable: as, in the same inscription, ESSENT, they might be, by the side of SVPERASES, thou mayest have conquered; but it must not be inferred that they were pronounced as single consonants.

Syllabic and Unsyllabic Function.

82. Whenever two or more sounds are combined in a syllable, one of them excels in acoustic prominence: as, a in English pat; n in the group pnd in opnd (opened); l in the group tld in bottld (bottled); and s in the group pst. This sound is said to have syllabic function or to be syllabic; in the examples given, a, n, l, and s are respectively syllabic. All the other members of each group are termed unsyllabic.

83. Vowels are almost always used in syllabic function. When, in rare cases, they are unsyllabic, this fact is usually indicated in phonetic works by an inverted half-circle, ̭, placed under the vowel; so in the case of diphthongs to indicate the subordinate member: as ai̭, oḙ, ṷi (49). Latin omnia and English glorious, when pronounced as words of two syllables, would be written omni̭a (2503), glori̭ous. When sounds other than vowels have, in rare cases, syllabic function, this fact is noted in phonetic works by a point,., or circle, ˳, under the letter: as, Latin *agṛs, *agr̥s (111, b), English opṇd, opn̥d.

Accent.

84. The relative force with which the different syllables of a word are uttered varies. Such variation in emphasis is called stress accentuation.

The degrees of stress are really infinite, but for practical purposes it is sufficient to distinguish between (1.) the strongest stress (chief accent); (2.) a weaker stress (secondary accent); and (3.) absence of stress (atonic syllable). In the English contradict, the last syllable has the chief accent, a secondary accent falls on the first, and the second syllable is unstressed.

85. It is not customary to indicate the place of accents in Latin by special signs. When, for special reasons, signs are used, ´ denotes the chief accent, ` the secondary accent, while the unstressed syllables are left unmarked.

The Classical Accent.

86. In classical Latin the place of the chief accent may be determined by the following rules.

(1.) Words of two syllables have the accent on the penult (175): as,

hómo; ā́cer.

(2.) Words of more than two syllables have the accent on the penult when that syllable is long (177); otherwise on the antepenult: as,

palū́ster, onústus (177); mulíebris, génetrīx (178); árborēs, árbutus, gladíolus.

87. A short penult retains the accent in the genitive and vocative with a single ī from stems in -io- (456, 459): as, genitive, cōnsílī; impérī; genitive or vocative, Vergílī; Valérī; Mercúrī. For calefácis, &c., see 394.

88. In a few words which have lost a syllable the accent is retained on the last syllable; such are (1.) compounds of the imperatives dīc and dūc (113): as, ēdū́c; (2.) nominatives of proper names in -ās and -īs for -ātis and -ītis: as, Arpīnā́s, for Arpīnā́tis; Laenā́s; Maecēnā́s; Quirī́s; Samnī́s; also nostrā́s, vostrā́s; (3.) words compounded with the abbreviated (113) enclitics -c for -ce and -n for -ne: as, illī́c; tantṓn; audīstī́n (for the shortening of the final syllable: as, vidén, dost see?, see 129); (4.) audī́t, contracted from audīvit (154, 893). The Latin grammarians prescribe the circumflex (90) for all these long syllables.

Earlier Recessive Accent.

89. In the preliterary period of the Latin language, the accent tended to go as far from the end of the word as possible (recessive accent). Thus, while the classical accentuation is inimī́cus, the older period accented ínimīcus. In literary Latin this early recessive accent has survived, only in Plautus’s accentuation of words of the form ⏑ ⏑ ⏑ ⏓ (proceleusmatic or fourth paeon, see 2521), in which he stresses the first syllable: as, fácilius (classical facílius); vóluerat (classical volúerat). But in many instances the early recessive accent may be traced in literary Latin by the phonetic changes which it produced (102 ff.).

90. Musical element. The native Latin grammarians slight the stress accentuation and pay much attention instead to the variations in pitch. But they are so greatly dependent on their Greek models that they are unsafe guides in this matter. It is, however, probable that a stressed vowel was uttered on a higher key (acute) than an unstressed vowel (grave), and that in certain syllables the long, accented vowel showed a rise and fall (circumflex): as, illîc (88).

91. The force of the Latin stress accent must have varied at different periods and in different localities, as it now varies in the Romance countries. The early recessive accent seems to have been fairly emphatic; but the stress in classical Latin was probably weak and the difference between accented and unaccented syllables was much less marked than it is in English.

Proclitics and Enclitics.

92. Proclitics are unaccented words which are pronounced as a part of the following word; they are: (1.) The relative and indefinite pronouns and their derivatives; (2.) Prepositions.

(a.) Thus, quō diē, pronounced quōdíē; quī vīxit, quīvī́xit; genus unde Latīnum, génus undeLatī́num. Similarly quamdíū, as long as; iamdíū, this long time. A distinction is thus made between the interrogative quālis (1526), which is accented, and the relative quālis (1831) which is proclitic (Quint. 1, 5, 26); cf. the English who, which is accented when interrogative, and proclitic when relative. (b.) circum lītora, pronounced circumlī́tora; ab ōrīs, pronounced abōrīs (Quint. 1, 5, 27); in inscriptions and manuscripts prepositions are often united in writing with the following word. Phrases like extemplō, suddenly, invicem (94), in turn, are commonly written and accented as one word. But the preposition is accented when it is followed by a monosyllabic unemphatic (and therefore enclitic) personal pronoun: as, ín mē; ábs tē (but abs tḗ, if tē is emphatic). All prepositions used as adverbs (1402) have an independent accent.

93. Enclitics are words which have no accent of their own, but are pronounced as a part of the word preceding. This increase of the number of syllables produced certain accentual changes, all the details of which are not clear. When the enclitic was monosyllabic the place of the accent seems to have been determined as in 86; thus vídēs, but vidḗsne; Látiō, but Latiṓque. Again, when by the addition of a monosyllabic enclitic the accent falls on the fourth syllable from the end, a secondary (84, 85) accent was probably placed on the penult: as, perī́cula, but perī́culàque. The Roman grammarians agree, however, in demanding that everywhere the syllable preceding the enclitics -que, -ne, -ve, and -ce should be accented. In deinde and subinde there is authority for placing the accent on the first syllable.

Enclitics are: (1.) Unemphatic personal and indefinite pronouns: as, in mē, pronounced, ínmē; dā mihi, dā́mihi; sīc tibi, sī́ctibi; sī quis, sī́quis; nē quid, nḗquid. (2.) Verbs when used as auxiliaries: as, possum for pót(e) sum (752); quī́ libet (2401); vidḗlicet, īlicet, scīlicet (712); quámvīs (1903); the forms of esse in compound tenses (719), so that est is frequently combined, even in writing, with the preceding past participle (747). (3.) The particles -ne (-n), -ve, and -ce (-c): as, satísne or shortened satín; Hyrcānīsve Arabī́sve; istī́ce or shortened istîc (90), adhûc (90). (4.) The copulative conjunction -que: as, Latiṓque, līmináque. (5.) The preposition cum when it follows (1435) its case. (6.) The particle quidem: as, sī quidem, síquidem (131). (7.) Other enclitics are: -met (650): as, egómet; -dem: as, ibídem; -nam: as, ubínam; -dum (1573): as, agédum; -inde: as, déinde, próinde (which are disyllabic in verse), and súbinde; -tum; as, etiámtum; -per: as, parúmper; the vocative when it was closely joined to the preceding word, e.g. an imperative: as, dī́c puer (106).

94. Two words expressing what is really one single idea are often bound together by one accent, one of them acting the part of either a proclitic or enclitic.

Thus, with the earlier recessive accent (89), Iū́piter (133; 389; originally a vocative which came to be used as nominative; for the change of pater to piter see 104); ínvicem, in turn; dḗnuō for dḗ nóvō (106); with the later, classical accent, lēgislā́tor, paterfamíliās, orbisterrā́rum, extémplō, imprī́mīs. When unemphatic ille and iste preceded their noun and had practically the value of our definite article they formed a unit with the following noun and thus the accent might fall on their last syllable: as, illé pater, isté canis. This use is particularly common in vulgar and late Latin (see 112).

Change of Sound.

(A.) Vowel Change.

CHANGE OF DIPHTHONGS IN ACCENTED SYLLABLES.

95. Of the six original diphthongs au, ou, eu, and ai, oi, ei, the only one which preserved its original sound in the classical period is au. Of the rest only ae (for older ai) and, in a few words, oe (for older oi) remained diphthongs; all the others had become monophthongs.

96. Change of ai. ai is common in inscriptions: as, AIDILIS, PRAITOR. Toward the end of the republic the two elements of the diphthong had been partially assimilated to ae (49): as, aedīlis (Quint. 1, 7, 18). This is its pronunciation in the classical period. Between 130 and 100 B.C. ai is displaced by ae in public documents; but the old-fashioned ai was often retained in private inscriptions. Still later the two elements completely converged to ē. In provincial Latin ē is found as early as 200 B.C.: as, CESVLA for CAESVLLA; in Rome itself the pronunciation ‘Cēcilius’ for Caecilius, and ‘prētor’ for praetor was derided as boorish; but by 71 A.D. ae was verging toward ē even in the court language: the coins of Vespasian have IVDEA as well as IVDAEA. In the 3d and 4th century A.D. ē became the prevalent sound.

97. Change of au. The diphthong au, which was preserved in educated speech, was changed to ō in rustic and colloquial pronunciation (see the anecdote related by Suetonius, Vesp. 22): as, cōpō, innkeeper, for caupō; plōstrum for plaustrum (barge), cart: Clōdius for Claudius. Some of these gained literary currency: as, cōdex, book, caudex, block; fōcāle, neckcloth, faucēs, throat. The form sōdēs (1572) for si̭ audēs = sī audēs (Cic. O. 154) is a colloquialism.

98. Change of ei. ei as a genuine diphthong is common in old inscriptions: as, SEI; SEIVE; ADEITVR; DEIXERVNT; FEIDA. In classical Latin it has passed into ī: thus, sī, if; sīve, either; adītur, is approached; dīxērunt, they said; fīda, faithful. An intermediate stage between the old diphthong ei and the classical ī was a very close (46) ē: as, PLOIRVME (465) for plūrimī; IOVRE (501, 507) for iūrī. For the orthographical use of ei as a spelling for the long ī-sound, see 29.

99. Change of oi. The development of oi was parallel to that of ai. It first passed into oe: as, COIRAVERVNT and COERAVERVNT, they cared; OITILE, useful, and OETI, to use; LOIDOS and LOEDOS, play,—all in old Latin. In classical Latin it has further been changed in accented syllables to ū: as, cūrāvērunt, ūtile, ūtī, lūdus. But oe was retained in classical Latin (1.) when a secondary diphthong (48), the result of contraction (120), and (2.) in a few words like foedus, treaty, perhaps as an archaizing, legal term; foedus, ugly; poena, penalty, perhaps through the influence of Greek ποινή (in the verb pūnīre, to punish, the regular ū appears); proelium, skirmish; foetor, stench; and moenia, walls, perhaps because there was a word mūnia, services. The connection of nōn, not, with noenum (455; 1444; 699) is difficult because of the unusual development of oe to o, for which the Praenestine form CORAVERONT is the only parallel.

100. Change of ou. ou, found in inscriptions down to about 90 B.C., passed, in classical Latin, into ū: as, POVBLICOM, NOVNTIATA, IOVDEX; later pūblicum, public, nūntiāta, notified, iūdex, judge.

101. Change of eu. Primitive (48) eu appears in classical Latin only in the interjections eu, heu, ē̆heu, heus. Every other original eu had, even in old Latin, passed into ou and developed like the latter: as, *neumen (Greek νεῦμα) became first *noumen, then (100) nūmen. With the exceptions noted above, the diphthong eu, as it appears in Latin, is always of secondary origin (48), the result of the two vowels e and u meeting in composition: as, neu, neither, from nē-ve; neutiquam, from nē and utiquam (124).

WEAKENING IN UNACCENTED SYLLABLES.

102. The vowel of an unstressed (atonic) syllable is often weakened, changing its quantity or quality or both. This is especially the case in syllables immediately preceded by the chief accent (posttonic syllables). The following changes took place at an early period when Latin still possessed the old, recessive accent (89).

WEAKENING OF SIMPLE VOWELS IN MEDIAL SYLLABLES.

103. (a.) Atonic medial e before a single consonant was weakened (with the exceptions given under b.) to i: as, cólligō, collect, from legō; óbsideō, besiege, from sedeō; cértāminis, of the contest, from certāmen (224); flāminis, from flāmen (470). And so probably hic (664) arose from *hec or *hoc (105, g) when used as proclitic (92). Before the labials p, b, f, and m this weakened sound was intermediate between i and u (28), and both spellings occur: as, quadripēs and quadrupēs, four-footed; alimentum, nourishment; monumentum, monument. The choice of i or u was probably governed by the quality of the stressed vowel in the preceding syllable: viz., u after o and u, and i after a, e, and i. But such distinction is only imperfectly maintained in classical Latin.

(b.) But before two consonants, before r, before vowels, and after i, atonic e does not change: as, lévāmentum (224), but lévāminis, of consolation; óbsessus (but óbsideō), possessed; sócietās, society, from the stem socie- (but nóvitās from the stem nove-); géneris, of the kind; ádeunt, they approach.

104. (c.) Atonic medial a, except in the cases mentioned below under (d.), (e.), and (f.), was first weakened to e and then underwent the same changes as atonic medial e (103): as (before single consonants), cṓnficiō, accomplish, from faciō; ī́nsiliō, jump in, from saliō (1019); rédditus, restored, from datus; trícipitem, three-headed, from *trícapitem (caput), Cic. O. 159; occiput, back of the head, and sinciput, jole (478). In compounds of iaciō (940), -iaciō is weakened in early Latin to -ieciō (as, conieciō, 940), but later to -iciō (as, subiciō). This last form may be due to syncope (111, a) of the radical a. The spelling -iiciō (as, subiiciō) is late and faulty (52). It does not occur in republican inscriptions and owes its origin to a confusion of the two forms conieciō and coniciō. (On the quantity of the vowel of the prepositions in these compounds of iaciō, see 122 e); (before p, b, f, m) áccipiō, accept, and óccupō, occupy, from capiō; cóntubernālis, room-mate, from taberna; ábripiō, to snatch away, from rapiō; (before two consonants) pépercī, I have spared, from parcō; áccentus, accent, from cantus; (before r) péperī, I brought forth, from pariō.

(d.) But an a in the preceding syllable may protect the atonic a: as, ádagiō, ádagium, proverb, but prṓdigium, miracle (144).

(e.) Atonic medial a before the guttural nasal (62) n followed by g changed to i (138): as, áttingō, touch, from tangō.

(f.) Atonic medial a before l followed by any consonant save l changed to u (both l and u being guttural, 60, 44): as, éxsultāre, to leap up, from saltāre; but féfellī, I deceived, from fallō.

105. (g.) Atonic medial o, when followed by a single consonant, first changed to e and then underwent all further changes of medial atonic e: as, hóminis, from *homon-is (485); ímāginis, for *imāgonis, 226 (nominative imāgō, 485); cúpīdinis, for *cupīdonis, 225, (nominative cupīdō, 485); vírginis, for *virgonis (nominative virgō, 470); ī́licō, from *in-slocō, on the spot (169, 4).

(h.) Before two consonants or before guttural l (60) atonic medial o changed to u: as, éuntis, from *éontis (Greek ἴοντος); sēdulō, from sē dolō (1417). But a preceding v or u protects o (107, c).

(i.) Before r, atonic medial o was retained: as, témporis, of time; except when u in the preceding syllable induced a change to u: as, fúlguris, of lightning (for the -r in the nominative singular fulgur instead of -s, see 154).

106. (k.) Medial -av-, -ov-, and -iv- in posttonic syllables were weakened to u: as, dḗnuō from dḗnovō (94); ábluō from ablavō. The form puer, boy, arose from the older POVER in enclitic vocatives (93, 7) and was thence transferred to the nominative like piter in Iūpiter (94).

WEAKENING OF SIMPLE VOWELS IN FINAL SYLLABLES.

107. (a.) In final syllables unaccented original e before s and t was weakened to i: as, salūtis, of safety, from older salūtes (507).

(b.) Final i became e: as, ante for *anti (Greek ἀντί and anti-cipāre); nominative singular mare, from the stem mari- (526).

(c.) In final syllables o before consonants changed to u except when preceded by u or v: as, fīlius, son, for old Latin fīlios (452); ferunt, they carry, for older feront; femur, thigh, nomin. sg. from the stem femor- (489); genus, kind, for *genos, Greek γένος; but vīvont, they live; salvom, safe. Not long before the beginning of our era o here also changed to u and appears to have coalesced with the preceding v (Quint. 1, 7, 26): as, in inscriptions: INGENVS (nomin. sg.) for ingenuos; SERVM, slave (acc. sg.), for servom; NOVM for novom, something new; so also boum, oxen (gen. pl.), for bovom (494). But inasmuch as the majority of forms in the paradigms of these words retained their v, it was restored in most cases, by analogy, to the forms which had lost it: as, servum for serum, because of servī, servō, etc.; vīvunt for vīunt, because of vīvō, vīvis, vīvit, etc.

(d.) When the stems fac- (facere, do), cap- (capere, take) appear as second members of compounds, their a changes in final syllables to e: as, artifex, artisan; auceps, bird-catcher. After the analogy of these words, compounds with dīcere and īre have e in the nom. sg.: as, iūdex, iūdicis, judge (from iūs and dīcere); comes, companion (from com, with, and īre); see 136, 2.

WEAKENING OF DIPHTHONGS IN UNACCENTED SYLLABLES.

108. Diphthongs, whether medial or final, are treated alike in atonic syllables.

(a.) Atonic ei, oi, and ai (ae) became ī: as, lupī, wolves (nom. pl.), for *lupoi (Gr. λύκοι); bellī, in war (loc. sg., 460, 1338), for *bellei (Greek οἴκει) or *belloi (Greek οἴκοι); éxīstimō, I consider, from aestimō; cóncīdō, I strike down, from caedō; Cicero, O. 159, mentions inīcum, unfair, for *ínaecum, and concīsum for *cóncaesum; so also, probably, hīc, this, arose from hoic (662) when used as a proclitic (92).

(b.) Atonic ou and au became ū: as, ínclūdō, I include, from claudō; áccūsāre, to accuse, from causa.

109. There are not a few cases in which the atonic vowel does not conform to the rules given above (102-108). These are usually compounds which show the vowel of the simple verb. Some of these were formed at a time when the early recessive accent was no longer in force and consequently there was no cause for weakening; in others the vowel of the simple verb was by analogy substituted for the weakened vowel of the compound: as, appetō, I strive after, from petō, which ought to have i like colligō, collect, from legō; intermedius, intermediate, but dīmidius, half; dēfraudāre, to cheat, by the side of dēfrūdāre from fraudāre; instead of the common redarguō, I refute, Scipio Africanus minor Pauli filius (185-129 B.C.) said rederguō, and pertīsum for pertaesum, but both Cicero (O. 159) and Lucilius discountenance pertīsum as the sign of a pedantic prig. In a few cases the reverse process took place, and the weakened vowel which arose in the compound was transferred to the simple verb: as, clūdō, I close (958), for claudō, which owes its ū to compounds like occlūdō. For a case where the vowel of the preceding syllable acted as a stay to the expected change, see 104, d.

LOSS IN UNACCENTED SYLLABLES.

110. Only vowels which are short and atonic may be lost. The loss of a medial vowel is called Syncope; of an initial vowel, Aphaeresis; of a final vowel Apocope.

111. Syncope. (a.) Loss of a posttonic vowel, entailing the loss of a syllable, occurs in ardus (Lucil.; for ă see 128) for the common āridus, dry; caldus by the side of calidus, warm (Quint. 1, 6, 19); reppulī, I pushed back, and rettulī, I carried back, stand for *répepulī and *rétetulī (861); pergō, I proceed, stands for *perregō from regō (cf. cor-rigō, ē-rigō, where the e is weakened, 103, and porrigō, porgō, where it is either weakened or lost), hence it forms its perfect perrēxī (953): pōnō, I place, is for *posnō (170, 2) from *po-sinō (112), hence it forms its past participle positus (972); for iūrgō, I blame. Plautus has iūrigō; *ūsūripō (from ūsus and rapere) yields ūsurpō, I utilize; *gāvideō, hence gāvīsus (801), gives gaudeo, I rejoice, converting āṷ to aṷ before the following d (128); in a similar way auceps, bird-catcher, is formed from *aviceps (avis, bird, and capere, catch); claudere, lock, from *clāvidere (clāvis, key); aetās, age, for áevitās (262); praecō, herald, for *práevicō (105, g) prae-vocō (211); also with change of ou to ū (100), prūdēns, prudent, for *proudēns from providēns, foreseeing; nūper, lately, from *noviper; nūntius, messenger, from *noventius (333); iūcundus, joyful, from iuvicundus (Cic. Fin. 2, 14). But forms like pōclum, cup, saeclum, age, do not belong here, as they are original and not derived by syncope from pōculum, saeculum; cf. 172.

(b.) Where, through the loss of a vowel, l or r would come to stand between two consonants, or where they would be final and preceded by a consonant, l and r become syllabic (83) and the syllable is thus maintained. Syllabic l is represented by ul, syllabic r by er (172, 3). The development of such intercalary vowels as u before l and e before r is called Anaptyxis (172). Thus, *sacri-dōts (cf. sacri-legium) became first *sacr̥dōts by syncope, then sacerdōs, priest, by anaptyxis; *ācribus (cf. ācri-mōnia, pungency) first became *ācr̥bus then ācerbus, pungent; *agrilos (267, cf. agri-cola, farmer) became first *agr̥los, then *agerlos, and finally, by assimilation of the r to l (166, 7), agellus, small field; from *dis-ficilter (adverb from dis- and facilis) arose *difficl̥ter and difficulter, with difficulty. The nominative sg. of the following words is to be explained thus. ager (451) was originally *agros (cf. Greek ἄγρος), which changed successively to *agr̥s, *agers, and ager (for the loss of -s see 171, 1 and 3). Similarly *ācris, passing through the stages of *ācr̥s, *ācers, became ācer (627), and *famlos by way of *faml̥s, *famuls, became famul (455), to which later the common ending of nouns of the o-declension was added, giving famulus.

112. Aphaeresis. Aphaeresis hardly occurs in literary Latin. In the pronoun iste the initial i is sometimes dropped (667); this loss implies an accented ultima (94). A trace of prehistoric aphaeresis is found in the prefix po- for *apo (Greek ἀπό) in pōnō, I place, for po-s(i)nō (111, a).

113. Apocope. Under the same conditions under which a medial vowel was syncopated, the final vowel of a word which stood in close union with the following word, as a preposition with its noun, was lost. In this way *peri (Greek περί) became per; *apo (Greek ἀπό) became ap, ab (164, 2); *eti (Greek ἔτι) became et. Similarly the final -e of the enclitics -ce, -ne, not, and -ne interrogative was lost: *sī-ce became sīc, so; *quī-ne, quīn, why not; habēsne, haben, hast thou; the imperatives dīc, say, dūc, lead, and fac, do, stand for earlier dīce, dūce, face (846); the shortened form em for eme (imperative of emere, take) has been turned into an interjection (1149). In the same way nec arose by the side of neque; ac by the side of atque (158). Final -e has also been dropped in the nominative sg. of a number of polysyllabic neuter stems in -āli and -āri (546): as, animal, animal, for *animāle, exemplar, pattern, for *exemplāre. See 536, 537. It must, however, be remembered that in most of the cases given the loss of a final vowel would also result from elision (119) before the initial vowel of the following word.

COMBINATION OF ADJACENT VOWELS.

114. Hiatus. A succession of two vowel sounds not making a diphthong is called Hiatus.

When in the formation of words by means of suffixes or prefixes or through the loss of an intervening consonant, two vowels come into contact within a word we speak of internal hiatus; the term external hiatus comprises those cases where, in connected discourse, the final vowel of one word comes into contact with the initial vowel of the following word. For the latter kind, see 2474.

115. The treatment of vowels in internal hiatus is four-fold: (1.) The hiatus may remain; (2.) the two vowels may be fused into one (Contraction); (3.) one of the two vowels may be dropped (Elision); and (4.) the two vowels may be combined into a diphthong.

116. Hiatus is maintained (a.) between two adjacent vowels the second of which is long and accented (according to the classical accentuation): as, coḗgi, I forced, and coā́ctus, forced (937); but cōgō (118, 3). For coepi, instead of coḗpī, I began, see 120.

(b.) In many prepositional compounds when the members were still felt to be independent: as, praeesse (the contracted form praesse is found in inscriptions); dēerunt, they will be wanting, by the side of dērunt; coalēscō, grow together (the contracted form cōlēscō appears in Varro); cooptāre, coöpt, cooperiō, I cover up (by the side of rare cōptāre, cōperīre); coïtus, meeting, by the side of coetus (120).

(c.) A comparatively large number of vowel combinations remain unchanged: as ea and eā in eam, her, and meā, by my (fem. sing.); ia and iā in māria, seas, viātōris, of the traveller; ua and uā in bēlua, monster, suā, through her (fem. sg.); iē in quiēs, quiet; uē in luēs, pestilence; eī in meī, of me; uī in tuī, of thee; eō in meō, by my (masc. sing.).

117. Synizesis. In these combinations the first vowel is sometimes made unsyllabic (83). This is called synizesis (2499) and is not rare in poets, being often the only means of adapting a word to the requirements of certain metres. Thus, fortuītus (- ⏑ - ⏓) must appear in a hexameter as fortvītus (fortṷītus). See 2499, 2503.

118. Contraction. (1.) Two like vowels may unite in one long vowel; rapidity of utterance was favourable to such fusion. In compounds, the desire to keep the members distinct often prevented it. So always nēmō, nobody, for *neemō from *ne-hemō, no man (for the loss of h, see 58, 150; for e in *hemō, see 144); and by the side of the open forms, nīl from nihil, nothing; vēmēns from vehemēns, rapid (connected with the verb vehō); rarely dērunt, they will be wanting, and dēsse, to be wanting, for dēerunt, dēesse; dēlēram, I had destroyed, from *dēlēeram for dēlēveram (for the loss of v, see 153), see 890; passūm, of paces, for passuum (591).

(2.) A diphthong absorbs the following vowel: as, praetor, older praitor, praetor, from *prai-itor, who goes before; inscriptions show praerunt for praeerunt, they will be before; for praebēre, to furnish, the open form praehibēre occurs in Plautus (1004).

(3.) If two unlike vowels are contracted at all, they usually unite in the long sound of the first vowel. Thus, o and a yield ō: as, cōgō, I force, from co-agō; cōgitō, I think, from co-agitō. Similarly Varro has cōlēscat, it may combine, for co-alēscat. o and e yield ō: as, prōmō, bring out, cōmō, put up, for pro-emō, co-emō (953). ē and a yield ē: as, dēgō, I pass away, from dē-agō (937). i and e in the termination of the vocative of -io- stems probably contracted to -ī; as fīlī from *fīlie, 459. But in denominative (365) and other verbs of the first conjugation ā and ō contract into ō: as, amō, I love, from *amā-ō (cf. Greek τιμά·ω); and ā and ē into ē: as, amēs, thou mayest love, for *amā-ēs.

119. Elision. Only rarely the first of two successive vowels is dropped: as, nūllus, no, for *ne-ūllus; likewise the final vowel of the first member of nominal compounds: as, multangulus, with many corners, for *multi-angulus (cf. multi-cavus, with many holes); flexanimus, heart-rending, for *flexi-animus (cf. flexi-pēs, with bent feet).

120. Combination into diphthongs. The union of two successive vowels into a diphthong is equally rare: o and i are combined to oi, oe, in coetus, meeting, by the side of the open form coïtus (116, b); the perfect coepī (812), I began, owes its diphthong oe to forms in which the e was short and unaccented, such as the rare present forms coepiō for có-ĕpiō (813); for coḗpi (813, 863) would have remained unchanged (116, a). neuter, with the accent on the e, was pronounced as three syllables, later eu became diphthongal; neutiquam with synizesis (117) of e. e and ī̆ sometimes contract to e͡i in rēi (601, 602) and in de͡inde, dēin in the classic poets.

LENGTHENING.

121. Compensative lengthening. When certain groups of consonants are simplified by the dropping of a consonant, its time is absorbed by a preceding short vowel, which thereby becomes long. This is called Compensation. In many cases compensative lengthening is due to the loss of a preliterary sonant s (170, 2): as,

cānus, gray, from *casnus (cf. cas-cus, very old). See for other cases of this lengthening, 170, 5, quīnī, for *quincnī; 170, 6, īgnōscō, for *in-gnōscō.

122. Induced lengthening. Before certain groups of consonants short vowels have a tendency to become long: as,

(a.) The prefixes in- and con- before s or f lengthened their vowels in classical Latin (Cic. O. 159): as, īnsānus, mad; īnfēlīx, unhappy; cōnsuēvit, he grew used to; cōnfēcit, he accomplished. Elsewhere also the vowel before ns and nf appears to have been lengthened: as, fōns, fountain; pēnsus, weighty (Gell. 9, 6); forēnsis, forensic; cēnsor, censor; mēnsa, table; mēnsis, mouth; Valēns; Clēmēns; the o of īnsons, guiltless, however, is marked as short by the grammarian Probus.

(b.) A similar lengthening of the vowel before nc followed by t or s appears: as, ūnctus, anointed, from unguō (Gell. 9, 6); iūnctus, joined, from iungō (954), coniūnx, spouse, genit. coniugis (472); quīnctus, fifth, whence quīntus (170, 4) and quīnque, five, derive their ī; sānctus, hallowed.

(c.) Spellings like sIgnvm, sign (well supported in inscriptions), and dIgne, worthily (less well supported) show that i was at times lengthened before gn. The grammarian Priscian demands this lengthening for all vowels preceding the ending -gnus, -gna, -gnum.

(d.) A lengthened vowel before r followed by a consonant is also certain for some words like ōrdō, order; fōrma, shape.

(e.) Some speakers appear to have lengthened the vowel of prepositions like con-, sub-, ob-, in the compounds of iaciō (104, c); as ōbicit. This practice, which is disapproved by Gellius (4, 17), probably arose from the transfer by analogy of the quantity of the first syllable in forms like conieciant (940) to that of the shortened form. In the same way the occasional spelling CÓNIV́NX, spouse, for coniūnx, may owe its long ō to the analogy of cōiunx, CÓIVGI (170, 6).

(f.) Many verb stems ending in -g have a long vowel in the past participle before the suffix -to-: as, tēctus, covered, from tego (916); tāctus, touched, from tangō (925); pāctus, fixed, from pangō (925); fīctus, moulded, from fingō (954); pīctus, painted, from pingō. The evidence for ā in maximus is very scanty: one case of A with the apex (29, 3) in a faulty inscription.

(g.) Of the induced lengthenings enumerated above, only those given in (a.) (b.) (f.) seem to have been universal in classical Latin. The rest appear to have been local peculiarities, which, while making inroads upon the literary language, never gained full recognition.

123. (1.) Analogical lengthening. In noun stems in -o the stem vowel is lengthened in the genitive plural -ōrum (449, 462), by analogy to the stems in -ā (435): as, servōrum, of slaves, like mēnsārum, of tables. For other cases see 122, e.

(2.) Metrical lengthening. On the lengthening of a vowel (or a syllable) under the influence of verse-ictus, see 2505.

SHORTENING.

124. A vowel originally long is regularly shortened in classical Latin before another vowel, even though an h intervene: as,

taceō, I am silent, from the stem tacē- (365); seorsum, apart, deorsum, downward, from sē(v)orsum, dē(v)orsum (153).

125. In simple words a diphthong occurs before a vowel only in one or two proper names: as, Gnaeus, Annaeus, in which it remains long, and in Greek words. But the diphthong ae of the prefix prae is sometimes shortened before a vowel: as, pra͝eacūtus; pra͝eeunt; pra͝ehibeō; hence prehendō for *prae-hendō. Sometimes it coalesces with a following vowel: as, pra͡e͡optāvīstī.

126. An increased tendency to shorten a long vowel before another vowel can be traced in the history of the language: thus, classical fuī, I was, for Plautus’s fūī (750); clueō, I am called, for Plautus’s clūeō; perfect pluit, it rained, for Varro’s plūit (cf. plūvit, 823, 947); pius, pious, for Ennius’s pīus; see also 765.

127. But even in classical Latin there are cases where a vowel before another vowel remains long: thus,

(1.) Regularly, the ī of fīō, I am made, except before -er-, as in fierem (788, 789).

(2.) In dīus, godly, for dīvus (153), and the old ablatives dīū, dīō, open sky (used only in the expression sub dīū, sub dīō, i.e. sub dīvō).

(3.) In the ending ēī of the genitive and dative sg. of stems in -ē- (601) when an i precedes: as, diēī, of a day, aciēī, of the battle line, but reī, of the thing, for older rēī.

(4.) It may be mentioned here that rēī is said to occur in verse 6 times (Plaut. G. 2, Lucr. G. 2, D. 2); reī 9 times (Plaut. G. 2, Ter. G. 4, D. 1, Juv. G. 1, Sulp. Apoll. G. 1); re͞i 27 times (Plaut. G. 2, D. 3, Enn. D. 1, Ter. G. 9, D. 8, Lucil. G. 1, D. 1, Lucr. G. 2). fidēī G. 3 times (Plaut., Enn., Lucr.); fideī 11 times (Enn. D. 1, Man. G. 2, D. 1, Sil. G. 4, D. 1, Juv. G. 2); fidēi 5 times (Ter. G. 1, D. 3, Hor. 1). ēī 35 times (Plaut. 18, Ter. 8, Lucr. 9); eī some 17 times (Plaut. 12, Ter. 2, German. 1, Ter. Maur. 2); ēi 23 times (Plaut. 11, Ter. 8, Lucil. 3, Cat. 1).

(5.) Gāius retains its ā before the vowel i: thus, Gāius (trisyllabic).

(6.) In the pronominal genitives in -ī̆us (618), the quantity of i varied. The older dramatists use ī; later, ī was shortened, but variations in its quantity seem to have continued until long after the end of the republic; Cicero, DO. 3, 183, measures illius; Quintilian 1, 5, 18 ūnīus; the grammarian Priscian prescribes -īus for all except alterius, which should always have i, and utrius, in which the i is common (30). In verse the i is often short, except in neutrīus; utriusque has always short i.

(7.) The penult is long in the endings -āī, -āīs, -ōī, -ōīs, and -ēī, -ēīs, from stems in -āio-, -ōio-, and -ēio- (458) or -iā- (437): as, Gāī, Bōī, Pōmpēī, plēbēī: Gāīs, Bōīs, Pompēīs, plēbēīs, Bāīs; aulāī, pictāī.

(8.) Dī̆ana has ĭ as often as ī. ohē has ŏ̄; ē̆heu has ĕ in comedy, otherwise ē.

(9.) In many Greek words a long vowel comes before another vowel; as, āēr, Aenēās, Mēdēa. But early importations from Greek followed the general rule and shortened the vowel: as, platĕa (πλατεῖα), balinĕum, balnĕum (βαλανεῖον).

128. A long vowel preceding unsyllabic i̭ or ṷ followed by a consonant is shortened: as, gaudeō for *gāudeō (cf. gāvīsus, 111); claudo for clāudō (cf. clāṷis, 111).

Similarly a long vowel (unless long by contraction: as, nūntius, 111, a, cōntiō) preceding a liquid or nasal followed by a consonant is shortened: as, syncopated ardus from āridus (111), habentem, from the stem habē-. For cases of induced lengthening of the vowel before n followed by certain consonants, see 122.

129. Iambic shortening. The law of iambic shortening (2470) produced a number of important changes: thus,

(1.) In old dramatic verse iambic words (⏑ –) often shorten the long vowel. The poets after Plautus and Terence preserve the long vowel.

(a.) Nouns; G. eri, boni, preti. D. cani, ero, malo. L. domi, heri. Ab. levi, manu, domo, bona, fide. Plural: N. fores, viri. D., Ab. bonis. Ac. foris, viros, bonas. (b.) Verbs: eo, volo, ago; ero, dabo; vides; loces; voles; dedi, dedin; roga, veni; later poets sometimes retain cave, vale, and vide. The vowel may also be shortened when -n (1503) is added and s is dropped before -n (170, 2): rogan, abin; viden is also retained by later poets.

(2.) In a few pyrrhic words (⏑ ⏑) in -i, which were originally iambic (⏑ –), the poets in all periods retained final -ī at pleasure: these are,

mihī̆, tibī̆, sibī̆; ibī̆, ubī̆; also alicubī̆. The i of bi is always short in nēcubi and sīcubi, and usually in ubinam, ubivīs, and ubicumque; ibidem is used by the dramatists, ibīdem in hexameter. ubīque has always ī.

130. The following instances show that this law operated in prose speech also:

(1.) In iambic words of the ā- declension (432) the final -ā of the nominative singular was shortened; hence *equā became equa, mare. From these iambic words short final -a spread so that all stems in -ā- shorten the final ā of the nom. sg. (434).

(2.) The final -a in the nominative plural of neuter nouns of the o- declension (446), which appears in trīgintā, thirty, was likewise shortened, first in iambic words like iuga, yokes, bona, goods, then everywhere (461).

(3.) This law explains the short final vowel in homo (2442) by the side of sermō (2437, c) and similar cases, like the adverbs modo, cito (2442), bene, male (2440). In the same way arose the short final o of the first person in conjugation (2443): as, volo, dabo, dīxero by the side of scrībō; so also viden for vidēn (129, 1; 170, 2).

(4.) Of imperatives only puta, used adverbially (2438, c), ave, have (805; Quint. i, 6, 21; but Martial scans havē) as a salutation and cave, used as an auxiliary (1711), show the short final vowel in classical Latin. Elsewhere the long vowel has been restored, as amā, monē (845).

(5.) According to this rule calēfaciō, malēdīcō changed to calefaciō, maledīcō.

131. A long final vowel is shortened when an enclitic is added to the word: as siquidem from sī; quoque from quō.

132. A long vowel is regularly shortened, in the classical period, before final -t and -m and, in words of more than one syllable, also before final r and l.

Thus, soror, sister, for Plautus’s sorōr, from the stem sorōr- (487); ūtar, I may use, for Plautus’s ūtār (cf. ūtāris); bacchanal for Plautus’s bacchanāl; animal, exemplar from the stems animāl- (530) and exemplār- (537); but the long vowel is retained in the monosyllables fūr, thief, sōl, sun; pōnēbat, he placed, for Plautus’s pōnēbāt (cf. pōnēbās); iūbet, he commanded, for Plautus’s iūbēt; eram, I was, but erās; rēxerim, I may have ruled, but rēxerīs (877); -um in the genitive plural of -o- stems is for -ūm (462); mēnsam, table, for *mēnsām from the stem mensā-; rem, from rē- (rēs), spem from spē- (spēs).

TRANSFER OF QUANTITY.

133. (1.) In a few cases the length of the vowel has been transferred to the following consonant, the length of which is then indicated by doubling it (81): as, littera for lītera, LEITERAS; Iuppiter for Iūpiter; parricīda for pāri-cīda, murder of a member of the same clan (*pāro-, member of a clan, Doric πᾶός, a relative); cuppa for cūpa, barrel. The legal formula sī pāret, if it appear, was vulgarly pronounced sī parret (Festus).

(2.) Since the doubled unsyllabic i (i̭) between vowels (23; 166, 9; 153, 2) is commonly written single, the vowel preceding it is often erroneously marked long: as, āiō wrongly for aiō, i.e. ai̭i̭ō, I say; māior wrongly for maior, i.e. mai̭i̭or, greater; pēior wrongly for peior, i.e. pei̭i̭or, worse; ēius, of him, cūius, of whom, hūius, of him, all wrongly for eius, cuius, huius i.e. ei̭i̭us, cui̭i̭us, huii̭us (153, 2). In all these words the first syllable was long but not the vowel.

VARIATIONS OF QUANTITY.

134. (1.) In some foreign proper names and in a very few Latin words the quantity of a vowel varied. Vergil has Sȳchaeus and Sychaeus within six verses; also Āsia and Asia, Lavīnium and Lāvīnius; so also glōmus (Lucr.), glomus (Hor.); cōturnīx (Plaut., Lucr.), coturnīx (Ov.).

(2.) Sometimes such variations in vowel quantity are only apparent: thus, the occasional long final -ē of the active infinitive (darē, prōmerē) has probably a different origin from the usual -ĕ. For metrical lengthening, see 2505.

QUANTITATIVE VOWEL GRADATION.

135. The same stem often shows a long vowel in some of its forms and a short vowel in others. In most cases these variations of quantity were not developed on Latin soil but inherited from a much earlier period. Such old inherited differences in vowel quantity are called quantitative vowel gradation.

(1.) Instances of this are prō for *prōd (149; cf. prōdesse) and pro- (Greek πρό); nē and ne- in nescius; the couples regō, I rule, and rēxī; vehō, I draw, vēxī; veniō, I come, vēnī, where the long vowel is characteristic of the perfect stem (862); vocō, I call, and vōx, voice; regō, I rule, and rēx, ruler; legō, I read, and lēx, bill; sedeō, I sit, and sēdēs, seat; fidēs, confidence, and fīdō, I trust; dux (cf. ducis), leader, and dūcō, I lead, where verb and noun are differentiated by the quantity of the root vowel; and many others.

(2.) Sometimes the reduction of the vowel in certain forms amounts to complete loss, as in the adverbial ending -is- in magis (346, 363) compared with the comparative suffix -ios, -iōs (Nom. -ior, Genit. -iōris); in the oblique cases of the stem carōn- (nomin. sg. carō, 497), where the suffix becomes -n- (545), genitive car-n-is; in the suffix -ter, which becomes -tr- in all cases but the nom. sg. (pater, patris, etc., 470, 487); in the feminine -tr-ī-c- to the suffix -tor-; but the nom. sing. Caecīlis (465) for Caecīlios is probably due to syncope.

QUALITATIVE VOWEL CHANGES.

136. (1.) i before an r which goes back to an earlier voiced s (154) was changed to e: as, cineris, of ashes, for *cinisis, from the stem cinis (491); Faleriī, for *Falisiī, cf. Falis-cus; (formed like Etrūria, for *Etrūsia, cf. Etrūs-cī).

(2.) In the nominative singular of compounds like iūdex, judge (from iūs and dīcere), comes, companion (from com, with, and īre, go), the i of the second member of the compounds is changed to e (470) after the analogy of words like artifex, artisan, etc. (107, d).

137. e before -gn- became i: as, īlignus, from the stem īlec- (cf. īlex).

138. e before the guttural nasal (62) followed by a guttural mute was changed to i: as, septingentī, from septem; singulī, from the stem sem- in semel (for the assimilation of m see 164, 3); obtingō (925), I attain, for *óbtengō (104, c) from *ob-tangō (104, e).

139. A similar change took place in the group -enl- which became first -inl- and then -ill-: as, *signilum, diminutive of sīgnum (for ī, see 122, c), first changed by syncope (111) from *signilum to *sign̥lum, then to *sigenlum (172, 3), then to *siginlum, and finally to sigillum.

140. o before nc became u: as, homunculus, manikin for *homonculus, from the stem homon- (485); nūncupāre, name, for *nōn-cupāre (nōn- for nōm- (164, 3) = syncopated nōmen); hunc, him, for *honc, from hom-ce (662).

141. o before l followed by any consonant save l was changed to u: as, cultus, tilled, for *coltus, from colere; multa, fine, for old Latin molta. But o before ll is retained: as, collis, hill.

142. e before guttural l (60) was changed to o: as, solvō, I undo, from *seluō (se-, as in se-cordia, luō = Greek λύω); culmen, top, for *celmen, from *cellō in ex-cellō; volō, I wish, for *velō; but e is preserved before dental l (60): as in velle, velim (773). Before l followed by any consonant save l this o changes to u (141): as, vult.

143. In a number of words, notably in voster, your, vorsus, turned, vortex, eddy, and votāre, forbid, the forms with o were replaced, about the second century B.C. by forms with e: as, vester, versus, vertex, vetāre (Quint. 1, 7, 25).

ASSIMILATION.

144. In a few cases a vowel is influenced by the vowel of a neighbouring syllable: as,

nisi, unless, for *nesi; iīs, for eīs, to them (671, 674); diī, diīs, gods, for deī, deīs (450); nihil, nothing, for *nehil; homō, man, for *hemō (cf. nēmō, from ne-hemō, 118); see also 104, d; 105, i.

QUALITATIVE VOWEL GRADATION.

145. The same stem often shows different vowels in different forms. In most of these cases this difference was inherited from a very early period and continued in the Latin. Such old inherited variation of the quality of the stem-vowel is called qualitative vowel gradation. The qualitative variations may be accompanied by quantitative changes (135).

Often the verb and the noun are thus distinguished by different vowels: as, tegō, I cover, and toga, a garment, toga; precor, I beg, and procus, suitor, cf. English to sing and a song, to bind, and a bond. The different tenses of some verbs show a like gradation: as, capiō, I take, cēpī; faciō, I make, fēcī, cf. English I sing, I sang; I bring, I brought. The same occurs in derivation: as doceō, I teach, by the side of decet; noceō, I harm, by the side of nex (nec-s). The two vowels which occur most frequently in such gradation are e and o: as in stems in -o-, domine, dominus (for dominos); as variable vowel (824); genos (genus, 107, c) in the nom. sg. by the side of *genes- in the oblique cases (gen. generis for *genesis, 154); honōs by the side of hones- in hones-tus; modus, measure, for *modos (originally a neuter -s- stem like genus (487, 491), but transferred later to the -o- declension), by the side of modes- in modes-tus, seemly. See 187.

(B.) Consonant Change.

146. In a number of words which belong more or less clearly to the stem of the pronoun quo- (681), cu- (157), the initial c has disappeared before u: as,

uter, which of the two, ubĭ, where, unde, whence (711). For the conjunction ut, utī, that, connection with this pronominal stem is much more doubtful. The c- appears in the compounds with sī and nē̆: as, sī-cubī (cf. sī-quidem, sī-quandō), sī-cunde, nē-cubi, ne-cunde, ne-cuter.

147. d varies in a few words with l: as old Latin dacruma, tear, for later lacrima; dingua, tongue, for later lingua; odor, smell, by the side of oleō, I smell.

148. Very rarely, before labials, final d of the preposition ad varies with r: as, old Latin arfuērunt, they were present, for later adfuērunt (2257); arvorsum, against, for advorsum. The only instances of this in classical Latin are arbiter, umpire, and arcēssō (970), I summon, which shows r before a guttural.

149. (1.) Final d after a long vowel disappeared in classical Latin: thus, in the ablative singular of -ā- and -o- stems (426), and in the ablative-accusative forms mēd, tēd, sēd (648). The prepositions prō and sē (1417) originally ended in -d which is still seen in prōdesse, be of advantage, prōd-īre, go forth; sēd-itiō, a going-apart, sedition. According to the grammarians, the negative haud preserved its d before vowels, but lost it before consonants (1450).

(2.) Late inscriptions confuse final -d and -t: as FECID (729), ALIVT for aliud. But in very old Latin -d in the third person singular seems to be the remnant of a secondary ending (cf. the Greek distinction of primary -ται and secondary -το).

150. In a number of words f varies dialectically with h. In some of these f appears to have been original, in others h: as, old Latin fordeum, barley, for classical hordeum; old Latin haba, bean, for classical faba. The word fīlum, thread, appears as *hīlum in nihil, nothing, for *ne-hīlum.

151. h being a weak sound (58) was often lost between two like vowels, especially in rapid utterance: as, nīl, nothing, prēndere, take, vēmēns, rapid, by the side of nihil, prehendere, vehemēns; and always nēmō, nobody, for *ne-hemō, no man.

152. In some words h between two vowels is not original, but goes back to a guttural aspirate gh. Before consonants this guttural appears: as, vehō, I draw, vectus (953) from a stem vegh-, trahō, I drag, tractus (953) from a stem tragh-.

153. (1.) v not infrequently disappeared between two like vowels: as, dītior, richer, for dīvitior; sīs (Cic. O. 154), for sī vīs (774); lātrīna, for lavatrīna; fīnīsse, for fīnīvisse; dēlēram, for dēlēveram; and later also in perfect forms in which the preceding and following vowel differed: as, amāsse, for amāvisse. The abbreviated forms of the perfects in -vī (890) were common in Cicero’s (O. 157) and Quintilian’s (1, 6, 17) time. v also disappeared before o in deorsum, seorsum.

(2.) Old and original unsyllabic i (82; 83) disappeared everywhere between vowels. Wherever unsyllabic i appears between vowels it represents double i̭i̭, and is the result of the assimilation of g to i̭ (166, 9), or d to i̭ (166, 9), or of the combination of two i̭’s: as in ei-i̭us, quoi̭-i̭us (eius, quoius = cuius, 688). See 23; 166, 9. In all these cases the first i̭ joined to the preceding vowel (83) formed with it a diphthong, and the syllable is thus long (133, 2).

(3.) The combinations of unsyllabic (83) ṷ with the vowel u and of unsyllabic i̭ with the vowel i were avoided in classical Latin; see 52.

(4.) In composition, unsyllabic (82) i̭ after a consonant became syllabic in quoniam, since, for quomi̭am (164, 5), and etiam, also, for eti̭am (both compounds with iam).

154. In early Latin s between two vowels was voiced (75), and in the fourth century B.C. this voiced s changed into r. According to Cicero (Fam. 9, 21, 2) L. Papīrius Crassus, consul in 336 B.C., changed his family name Papīsius to Papīrius. Old inscriptions show frequently s for r: as, ASA, altar, AVSELII. This change of intervocalic s to r plays an important part in declension, conjugation, and derivation: as,

Nominative iūs, right, genitive iūris; spērō, I hope, derived from spēs; nefārius, wicked, from nefās; gerō, I carry, from a stem ges- which appears in ges-sī, ges-tus (953); erō, I shall be, from the stem es- in esse; the subjunctive ending -sem in es-sem appears as -rem after vowels: as, stārem; the infinitive ending (894, 895) -se in es-se appears as -re after vowels: as, legere, for *legese, to read, stāre, for *stāse, to stand. Where all oblique cases show -r- and only the nominative singular -s, the latter is sometimes changed to -r by analogy: as, arbor, tree, honor, honour, for original arbōs, honōs, by analogy to the oblique cases arboris, arborī, honōris, honōrī, etc. (487, 488). The final -s of the prefix dis- follows this rule: as, dir-imō, I take apart, for *dis-emō; but an initial s- of the second member of a compound remains unchanged: as, dē-sinō, I stop.

155. Wherever intervocalic s is found in classical Latin it is not original, but the result (a.) of earlier -ns-: as, formōsus, handsome, for formōnsus (63); (b.) of earlier -ss- (170, 7): as, ūsus for *ūssus, use (159); causa, thing, for caussa (Quint. 1, 7, 20); or (c.) it occurs in borrowed words like asinus, ass. (d.) There are a few words in which an r in a neighbouring syllable seems to have prevented the change: as miser, miserable (173).

156. Before the o described in 142 qu changed to c: as, incola, inhabitant, for *inquola, from *inquela; the stem quel- appears in in-quil-īnus, lodger.

157. As v before u (107, c), so qu was not tolerated before u, but changed to c.

Hence when, about the beginning of our era, the o of quom, when, sequontur, they followed, changed to u (107, c), they became cum, secuntur; thus equos but ecus, horse (452); reliquom but RELICVM, the rest; loquor, I speak, but locūtus (978). Much later, in the second century of our era, the grammarians restored the qu before u by analogy to those forms in the paradigm in which qu came before other vowels: as, sequuntur for secuntur by analogy to sequor, sequeris, sequitur, sequimur, sequimini, etc.; equus, equum, for ecus, ecum, by analogy to equī, equō, eque, equōrum, equīs, equōs.

158. qu before consonants or when final changed to c: as, relictus from the stem liqu-, leave (present, linquō, 938); ac, and, for *atc, by apocope from atque; nec, nor, by apocope from neque. See also *torctus (170, 3), quīnctus (170, 4).

159. When in the process of early word formation a t was followed by another t, the combination tt, unless followed by r, changed to ss: as, obsessus, besieged, sat upon, for *obsettus, from *obsed-tus (cf. sedeō). After long vowels, nasals, and liquids this double ss was simplified to s (170, 7): as, ūsus from *ūt-tus, used (cf. ūtor); scānsus, climbed, from *scant-tus for *scandtus (cf. scandō).

In this way arose a suffix -sus (906, 912) for the past participle of verbs ending in a dental, and this spread to other verbs (912): as mānsus, stayed, from maneō (1000), pulsus, pushed, from pellō (932). The regular participles of these two verbs still appear in the derivative verbs mantāre and pultāre, which presuppose the past participles *mantus and *pultus (371). If the double tt was followed by r it changed to st: as, assestrīx from *assettrīx, while *assettor changed to assessor.

160. But wherever the combination tt arose in historical times it remained unchanged: as, attineō; cette, syncopated for cé-d(i)te, i.e. the particle ce (93, 3) which is here proclitic, and the imperative date, give.

161. Initial dv (dṷ) changed to b, unless the v (ṷ) was converted into the corresponding vowel: as, bis, twice, for *dṷis (cf. duo); bidēns for *dṷidens, by the side of old Latin duidēns with vocalic u; bonus, good, for dṷonus, by the side of trisyllabic duonus; bellum, war, for *dṷellum, by the side of duellum with vocalic u; bēs, two thirds, for *dṷēs (2427). Cicero (O. 153) notes that the change of duellum to bellum affected even the proper name Duellius (name of the admiral who won the naval victory over the Carthaginians in 260 B.C.) which was changed to Bellius. Plautus always scans dṷellum disyllabic with synizesis (2503).

CHANGES OF CONSONANT GROUPS.

162. Many groups of consonants undergo changes in order to facilitate their pronunciation in rapid speech. These changes involve (a.) Assimilation of consonants; (b.) the development of consonantal glides; (c.) the loss of one member of the group; and (d.) the development of a vowel between the consonants.

ASSIMILATION.

163. Of two successive consonants belonging to different syllables (175), the first is, as a rule, assimilated to the second (regressive assimilation), rarely the second to the first (progressive assimilation). A consonant may be assimilated, either entirely or partially, to another consonant.

Assimilation is very common in prepositions prefixed to a verb.

164. Partial assimilation. (1.) A voiced mute before an unvoiced consonant became unvoiced: as, rēx, king, for *rēgs (cf. rēgis); rēxī, I guided, for *rēgsī (cf. regō); rēctus, guided, for *rēgtus; scrīpsī, I wrote, for *scrībsī (cf. scribō); scrīptus, written, for *scribtus; trāxī, I dragged, for *trāghsī; tractus, dragged, for *traghtus (152). The spelling did not always conform to this pronunciation: as, urbs, city, pronounced urps (54) but spelled with b by analogy to the oblique cases urbis, urbem, etc.; obtineō, I get, pronounced optineō.

(2.) An unvoiced mute before a voiced consonant became voiced. The prepositions ob, ab, sub, for *op, *ap, *sup, owe their final b to their frequent position before voiced mutes: as, obdūcō, abdīcō, sub dīvō. The forms *op (still preserved in op-eriō, I close, 1019) *ap (preserved in ap-erio, I open, 1019; cf. Greek ἀπό) and *sup (preserved in the adjective supīnus, supine) were then crowded out by ob, ab, and sub.

(3.) Nasals changed their place of articulation to that of the following consonant. Thus, dental n before the labials p and b became labial m: as, imbibō, I drink in, impendeō, I hang over. Labial m before the gutturals c and g became guttural n (62): as, prīnceps, leader, singulī, severally (the original labials appear in prīmus, semel (138)); hunc for *homce (662). Labial m before the dentals t, d, s became dental n: as, cōnsecrō, I consecrate, from com (cum) and sacrō; tantus, so great, from tam; quondam, once, from quom; tandem, at length, from tam. But sometimes the etymological spelling was retained: as, quamdiū, as long as. But m does not change to n before t or s in the inflection of verbs and nouns, where mt, ms develop into mpt, mps (167): as, sūmptus, sūmpsī, from sūmō.

(4.) p and b before n changed to m: as, somnus, sleep, for *sop-nus (cf. sopor); omnis, all, for *op-nis (cf. opēs); Samnium, for *Sabnium (cf. Sabīnī).

(5.) m before unsyllabic i (i̭) became n: as, quoniam (with vocalic i; 153, 4), since, for *quoni̭am from quom iam (1882); coniungō, I join together, for *comiungō.

(6.) c between n and l, and before m, changed to g: as, angulus, corner, with anaptyctical (172) vowel u for *anglus, from *anclus (cf. ancus); segmentum, section, from the stem sec- in secāre.

165. It appears that at a very early period the neighbourhood of a nasal changed an unvoiced mute into a voiced one: as, ē-mungō, I clean out, by the side of mūcus; pangō, I fix, by the side of pāc- in pāx, peace (gen. pāc-is).

166. Entire assimilation. (1.) One mute is assimilated to another: thus p or b to c: as, suc-currō, I assist; t or d to c: as, sic-cus, dry (cf. sit-is, thirst), accipiō, I accept; d to g: as, agglūtinō, I glue on; t or d to qu: as, quicquam, anything; t or d to p: as, appellō, I call; quippe, why? (1690).

(2.) A mute is assimilated to a spirant: thus, p to f in officīna, workshop, for *opficīna, syncopated form of *opificīna; d to f: as, afferō, I bring hither; when t is thus assimilated to s the result is ss after a short vowel, and s after a long vowel (170, 7) or when final (171); as, in the -s- perfects, concussī, I shook, for *concutsī (concutiō, 961); messuī, I mowed, for *metsuī (metō, 835); suāsī, I advised, for *suātsī (suādeō, 1000); clausī, I shut, for *clautsī (claudō, 958); haesī, I stuck, for haes-sī (868) from haerēre, stem haes- (154); in the same way possum, I can, for *potsum (cf. pot-est, 752); prōsum, I am of advantage, for *prōtsum (cf. prōd-esse); legēns, reading, for *legents (from the stem legent-, cf. genitive legent-is). An s is never assimilated to a following t: as, haustus, drained (1014), from the stem haus-, present hauriō (154). Forms like the rare hausūrus (Verg.) are made after the analogy of dental stems.

(3.) One spirant, s, is assimilated to another, f: as, difficilis, difficult, differō, I am unlike, from dis and facilis, ferō.

(4.) A mute is assimilated to a nasal: thus d to m in mamma, woman’s breast, from the stem mad- (cf. madeō, 1006); rāmus, branch, rāmentum, splinter, from the stem rād- (cf. rādō, 958) with simplification of the double m after the long vowel. d to n in mercēnārius, hireling, from the stem mercēd-, reward, (for mercennarius, see 133, 1); p to m in summus, highest, from the stem sup- (cf. super). A progressive assimilation of nd to nn belongs to the Oscan dialect, and occurs only very rarely in Latin: as, tennitur (Ter.), distennite (Plaut.) See 924; 950.

(5.) One nasal, n, is assimilated to another, m: as immōtus, unmoved. But an m before n is never assimilated: as, amnis, river.

(6.) Mutes or nasals are assimilated to liquids; thus n to l: as, homullus, manikin, for *homon-lus (cf. homun-culus); ūllus (274); d to l: as, sella, seat, for *sed-la from the stem sed- (cf. sedeō); caelum, chisel, from the stem caed- (cf. caedō) with simplification of the double l after the diphthong (170, 7); n to r: as, irruō, I rush in; and with progressive assimilation n to a preceding l: as, tollō, I lift, for *tolnō (833); fallō, I cheat (932); pellō, I push (932). But no assimilation is to be assumed for parricīda, which does not stand for patricīda (133, 1).

(7.) One liquid, r, is assimilated to another, l: as, pelliciō, I lead astray (956), for *per-liciō; agellus, small field, for *agerlos; pūllus, clean, from *pūrlos (cf. pūrus, clean).

(8.) A spirant, s, is assimilated to a preceding liquid in velle, wish, for *velse, ferre, carry, for *ferse (the infinitive ending -se appears in es-se, 895); facillimus, easiest, for *facilsimus (345); sacerrimus, holiest, for *sacersimus (344). But where ls and rs are not original but the result of lightening (170, 3; 10) they remain unchanged: as, arsī, I burnt, for *artsī from the stem ard- (cf. ardeō, 1000); alsī, I felt cold, for *alcsī from the stem alg- (cf. algeō, 1000).

(9.) g and d were assimilated to a following unsyllabic i (i̭) the result being (153, 2) ii (i̭i̭); thus peiior, worse, for *ped-i̭or, from the stem ped- (532), whence also the superlative pessimus for *petsimus (166, 2); maiior, greater, for *mag-i̭or (the stem mag- appears in magis); aiiō, I say, for *ag-i̭ō (the stem ag- appears in ad-ag-ium, prōd-ig-ium, 219). These forms were pronounced by Cicero with doubled i̭ (23), and traces of the spelling with double ii are still found (23), though in common practice only one i is written (153, 2). On the confusion of syllabic quantity with vowel quantity in these words, see 133, 2.

CONSONANTAL GLIDES.

167. Pronunciation of two successive consonants is sometimes facilitated by the insertion of a consonant which serves as a glide. Such insertion is not frequent.

In inflection a p was thus developed between m and s, between m and l, and between m and t (elsewhere mt changed to nt, see 164, 3): as, sūmpsī, I took, sūmptus, taken, from sūmere for *sūmsī, *sūmtus; and in the corresponding forms of cōmō, dēmō, prōmō (953); exemplum, pattern, for *exemlum from the stem em-, take (cf. eximere, 103, a).

DISAPPEARANCE.

168. A word may be lightened by the disappearance of an initial, a medial, or a final consonant.

Disappearance of an initial consonant is sometimes called Aphaeresis, of a medial, Syncope, of a final, Apocope.

169. Initial disappearance. (1.) Initial tl changed to l: as, lātus, borne, for *tlātus from tollō (187, 917).

(2.) Initial gn changed to n: as, nātus, born, for earlier GNATVS from the stem gen-, gnā (187); nōscō, I find out, for gnōscō, GNOSCIER (897); nārus, knowing, for the more frequent gnārus, nāvus, active, for gnāvus. Cf. the compounds cō-gnātus, cō-gnōscō, ī-gnārus, ī-gnāvus (170, 6) which preserve the g. But Gnaeus retained its G.

(3.) Initial d when followed by consonant i (i̭), disappeared: as, Iovis, Iūpiter, for *Di̭ovis, *Di̭ūpiter. Where the i was vocalic, d was retained: as, dīus.

(4.) Initial stl- first changed to sl and then to l: as, Old Latin stlocus, place, stlīs, law-suit (Quint. 1, 4, 16), STLOC, SLIS, classical locus, līs; also lātus, wide, for *stlātus. That a form *slocus existed is proved by īlicō (698, 703) from *in-slocō, on the spot (170, 2).

170. Medial Disappearance. (1.) c, g, p, and b disappear before s followed by an unvoiced consonant: as, sescentī, six hundred, for *sexcentī from sex; illūstris, resplendent, for *illūcstris from lūceō; discō, I learn, from *dicscō for *di-tc-scō (834), a reduplicated present from the root dec- (cf. decet) like gignō (from the root gen-), and sīdō (for *si-sd-ō, 170, 2, from the root sed-, 829). Sometimes prepositions follow this rule: as, asportō, I carry off, for *absportō, suscipiō, I undertake, for *subscipiō (subs formed from sub like abs from ab; sub-cipiō gives succipiō); occasionally also ecferō, for exferō, I carry out. But more frequently prepositional compounds remain unchanged: as, obscūrus, dark; abscēdō, I withdraw. In some words the lost consonant has been restored by analogy: as, sextus, sixth, for *sestus (cf. Sēstius) after sex; textor, weaver, for *testor after texō.

(2.) s before voiced consonants was voiced (75) and is dropped. If a consonant precedes the s this is dropped also. In either case the preceding vowel is lengthened. Voiced s alone is dropped: as, prīmus, first, for *prīs-mus (cf. prīs-cus); cānus, gray, for *casnus (cf. cas-cus); adverb pōne, behind, for *posne (cf. pos, 1410); dīlābī, glide apart, for *dislābi; īdem, the same, for ISDEM (678); iūdex, judge, for iūsdex; trēdecim, thirteen, for *trēsdecim. And with subsequent shortening of the final syllable (130, 3) abin, goest thou? for abisn(e); viden, seest thou? for vidēsn(e). Voiced s with the preceding consonant is dropped: as, trādūcō, I lead across, trānō, I swim across, for trānsdūcō, trānsnō; but in these prepositional compounds the -ns was often retained: as, trānsmittō, I send across; sēnī, six each, for *secsnī; sēmēnstris, every six months, for secsmēnstris; sēvirī, the Board of Six, for secsvirī; āla, wing, for *acsla (cf. ax-illa, Cic. O. 153); māvolō (779) for magsvolō from magisvolō, 396; tōles (plural), goiter, for *tōnsles (cf. tōnsillae, tonsils); pīlum, pestle, for *pīnslum from pīnsere, crush; two consonants and voiced s are dropped in scāla, stair, for *scand-sla (cf. scandō).

(3.) c falls away when it stands between a liquid and t, s, m, or n: as, ultus, avenged, for *ulctus from ulc-iscor (980); mulsī for *mulcsī from both mulgeō, I milk, and mulceō, I stroke; similarly other stems in -c and -g (1000, 1014); quernus, oaken, for *quercnus from quercus; tortus, turned, for *torctus from torqueō (for the change of qu to c, see 158); for fortis, brave, forctis is found in old Latin.

(4.) c drops out when it stands between n and t: as, quīntus, fifth, for older quīnctus (2412), from quīnque (for the change of qu to c, see 158; for the long ī in quīnque, see 122, b). But verbs having stems in -nc or -ng retain the c in their past participles: as, vīnctus, bound, from vincīre (1014); iūnctus, joined, from iungere (954). In pāstus (965) c has dropped out between s and t.

(5.) The group -ncn- was simplified to simple -n-, and the preceding vowel was lengthened: as, quīnī, five each, for *quīnc-nī (317); cō-nīveō, wink and blink, for con-cnīveō.

(6.) n before gn was dropped and the preceding vowel lengthened: as, ī-gnōscō, I forgive, for *in-gnōscō, cō-gnōscō, I know, for *con-gnōscō. In this manner (170, 5; 6) arises a form cō- by the side of con- (122, e): as, cō-nectō, cō-nubium, cō-ligātus (Gell. 2, 17, 8).

(7.) In the imperial age, ss after long vowels and diphthongs was regularly changed to s: as, clausī, I closed; ūsus, used (166, 2); but always ēsse, to eat (769); ll changed to l after diphthongs: as, caelum, chisel (166, 6); also when preceded by ī and followed by i: as, vīlla, country-place, but vīlicus (adject.); mille, thousand, but mīlia (642). Elsewhere ll was retained after long vowels: as, pūllus (166, 7), clean; rāllum, ploughshare, from rādō with suffix -lo- (209). In Cicero’s time (Quint. 1, 7, 20) the spelling was still caussa (155, b), matter; cāssus (930), fallen; divīssiō (cf. 912), division. Vergil also, according to Quintilian, retained the doubled consonants, and the best manuscripts of both Vergil and Plautus frequently show ll and ss for later l and s, as do inscriptions: as, PROMEISSERIT, he might have promised (49 B.C.); ACCVSSASSE, to have accused.

(8.) After a long vowel d was dropped before consonant u (v): as, svāvis, sweet, for *svādvis from svād- (cf. svādeō).

(9.) r before st was dropped: as, tostus, roasted (1004) for *torstus from the stem tors- (cf. torreo with assimilated -rs-, 166, 8).

(10.) -rts- changed to -rs: as, arsī, I burnt, for *artsī (1000). -rcsc- changed to -sc-: as, poscō, I demand, for *porcscō (834).

(11.) In ipse, self, for *is-pse, an s has disappeared before -ps-

(12.) (12.) d (t) disappears between r and c: as, cor-culum for cord(i)-culum (275).

171. Final Disappearance. (1.) A word never ends in a doubled consonant: as, es for *es-s, thou art, which Plautus and Terence still scan as a long syllable; and the following cases of assimilation: ter for *terr from *ters (cf. terr-uncius, a quarter of an ās, a farthing, 1272, for *ters-uncius, 166, 8); fār, spelt, for *farr, from *fars (489); fel, gall, for *fell, from *fels (482); in mīles, soldier, for *mīless from *mīlets (cf. Gen. mīlitis, 477) the final syllable is still long in Plautus. hoc, this, for *hocc from *hod-c(e)(the neuter *hod from the stem ho-, as istud, illud (107, c) from isto-, illo-) counts as a long syllable even in classical poetry.

(2.) No Latin word can end in two explosives: thus, final t is dropped in lac, milk (478); final d in cor, heart (476).

(3.) When final s was preceded by r or l, it was assimilated to these liquids, and final rr and ll were then simplified to r and l. See the examples under (1). Wherever final -rs and -ls appear they are not original but the result of the disappearance of an intervening consonant: as, puls, pottage, for *pults (533); pars, part, for *parts (533); all with syncope (111) of the vowel i in the nominative sg.

(4.) Original final ns was changed to s and the preceding vowel was lengthened: as, sanguīs, blood (2452), for *sanguins from the stem sanguin- (486). Wherever final -ns appears it is not original but the result of the disappearance of an intervening consonant: as, ferēns, carrying, for *ferents, from the stem ferent-; frōns, foliage, for *fronds, from the stem frond-.

(5.) A dental mute before final s is dropped: as, hērēs, heir, for *hērēds (475); virtūs, virtue, for *virtūts (477); nox, night, for *nocts (533); a labial or guttural mute is retained: as, fornāx (x = cs), furnace, from the stem fornāc- (531); lēx, law, from the stem leg- (472); urbs, city, from the stem urb- (480); ops from the stem op-, help (480).

DEVELOPMENT OF AN ANAPTYCTICAL VOWEL.

172. Certain consonant groups, notably those containing a liquid, are sometimes eased by the insertion of a vowel which develops between the consonants. This is called Anaptyxis (Greek ἀναπτύσσειν, unfold). It is the opposite of syncope of vowels (110, 111).

(1.) The suffix -clo- (242), changed to -culo-, being thus no longer distinguishable from the diminutive suffix -culo- (267): as, pōculum, cup, for pōclum (Plaut.); vehiculum, carriage, for vehiclum (Plaut.). But -clo- is more common in Plautus than -culo-, especially after long vowels. The suffixes -blo- (245), and -bli- (294) always show the anaptyctical vowel. Its colour depends on the nature of the l (60): as, stabulum, resting-place; stabilis, steady. The group -ngl- also changes to -ngul-: as, angulus (164, 6).

(2.) In words borrowed from the Greek an unfamiliar sequence of consonants was so lightened; as, mina, mina, for *mna (μνᾶ); and in Old Latin drachuma (Plaut.) for later drachma, drachma (δραχμή); techina, trick, from Greek τέχνη; Tecumēssa for Tecmēssa (Τέκμησσα).

(3.) Before syllabic (83) l and r a vowel is developed (111, b): as, íncertus, uncertain, for *íncr̥tus; fácultās, capability, for fácl̥tās. Likewise before syllabic n (139).

DISSIMILATION.

173. (1.) To avoid the repetition of the same liquid in successive syllables l is sometimes changed to r: as, caeruleus, sky-blue, for *caeluleus, from caelum; Parīlia, by the side of Palīlia, from Palēs; the suffix -clo- appears as -cro- after an l: as, lavācrum, bath, simulācrum, image (241); the suffix -āli- under like conditions changes to -āri-; as, molāre, of a mill (313), but augurāle, of an augur.

(2.) In a few cases repetition is avoided by dropping the sound once: as, praestīgiae, jugglery, for praestrīgiae. This also applies to the spirant s followed by a consonant, a combination which is not tolerated in successive syllables: as in the reduplicated perfects stetī, for *stestī; spopondī, for *spospondī (859), where the second syllable, and in quisquiliae, sweepings, for *squisquiliae, where the first syllable was lightened.

CHANGES WITHIN COMPOUNDS.

174. The final syllable of the first member of compounds (181) sometimes undergoes certain changes by analogy to other compounds:

(1.) The final -ā of ā-stems, by analogy to the more frequent -o-stems, usually changed to -o, which in atonic syllables became -i (105): as, āli-ger, winged, for *ālo-ger from ālā-.

(2.) Stems in -on- substitute -o- for -on- by analogy to the -o-stems: as, homi-cída, murderer, for *homo-cīda (105) from homon- (Nom. homō).

(3.) Some stems in -s substitute -o- by analogy to the -o-stems: as, foedi-fragus, treaty-breaking, for *foedo-fragus from the stem foedos- (Nom. foedus, Gen. foederis; 154).

Syllables.

175. A word has as many syllables as it has separate vowels or diphthongs. The last syllable is called the Ultima; the last syllable but one is called the Penult; the last syllable but two is called the Antepenult.

176. The quantity of single sounds (e.g. the quantity of a vowel) must be carefully distinguished from the quantity of the group of sounds or the syllable of which the single sound forms a part.

Length of Syllables.

177. A syllable is long if its vowel is long, or if its vowel is followed by two consonants or by x or z: as,

dūcēbās; volvunt. In dūcēbās both the vowels and the syllables are long; in volvunt the vowels are short, but the syllables are long; in cases like the last the syllables (not the vowels) are said to be long by position. h does not count as a consonant (58) and qu (or qv, 27) has the value of a single consonant only: thus, in adhūc and aqua the first syllable is short.

178. In prose or old dramatic verse a syllable with a short vowel before a mute or f followed by l or r is not long: as tenebrae. In other verse, however, such syllables are sometimes regarded as long. In compounds such syllables are long in any verse: as obruit.

Loss of Syllables.

179. The first of two successive syllables which begin with the same sound is sometimes lost. This is called Haplology.

Thus, sēmodius for sēmimodius, half a bushel; calamitōsus for *calamitātōsus, from the stem calamitāt- (262) and suffix -oso- (336); voluntārius, for voluntātārius (262, 309); cōnsuētūdō, for cōnsuētitūdō (264). See also 255; 379.