Latin Language

(i.) Back-palatal and Velar Sounds.—In point of its treatment of the Indo-European back-palatal and velar sounds, it belongs to the western or centum group, the name of which is, of course, taken from Latin; that is to say, like German, Celtic and Greek, it did not sibilate original k and g, which in Indo-Iranian, Armenian, Slavonic and Albanian have been converted into various types of sibilants (Ind.-Eur.* kṃtom = Lat. centum, Gr. (ἑ)-κατόν, Welsh cant, Eng. hund-(red), but Sans. ṡatam, Zend satƏm); but, on the other hand, in company with just the same three western groups, and in contrast to the eastern, the Italic languages labialized the original velars (Ind.-Eur. * qod = Lat. quod, Osc. pod, Gr. ποδ-(απός), Welsh pwy, Eng. what, but Sans. kás, “who?”).

(ii.) Indo-European Aspirates.—Like Greek and Sanskrit, but in contrast to all the other groups (even to Zend and Armenian), the Italic group largely preserves a distinction between the Indo-European mediae aspiratae and mediae (e.g. between Ind.-Eur. dh and d, the former when initial becoming initially regularly Lat. f as in Lat. fēc-ī [cf. Umb. feia, “faciat”], beside Gr. ἔ-θηκ-α [cf. Sans. da-dhā-ti, “he places”], the latter simply d as in domus, Gr. δόμος). But the aspiratae, even where thus distinctly treated in Italic, became fricatives, not pure aspirates, a character which they only retained in Greek and Sanskrit.

(iii.) Indo-European ŏ.—With Greek and Celtic, Latin preserved the Indo-European ŏ, which in the more northerly groups (Germanic, Balto-Slavonic), and also in Indo-Iranian, and, curiously, in Messapian, was confused with ă. The name for olive-oil, which spread with the use of this commodity from Greek (ἔλαιϝον) to Italic speakers and thence to the north, becoming by regular changes (see below) in Latin first *ólaivom, then *óleivom, and then taken into Gothic and becoming alēv, leaving its parent form to change further (not later than 100 B.C.) in Latin to oleum, is a particularly important example, because (a) of the chronological limits which are implied, however roughly, in the process just described, and (b) of the close association in time of the change of o to a with the earlier stages of the “sound-shifting” (of the Indo-European plosives and aspirates) in German; see Kretschmer, Einleit. p. 116, and the authorities he cites.

(iv.) Accentuation.—One marked innovation common to the western groups as compared with what Greek and Sanskrit show to have been an earlier feature of the Indo-European parent speech was the development of a strong expiratory (sometimes called stress) accent upon the first syllable of all words. This appears early in the history of Italic, Celtic, Lettish (probably, and at a still later period) in Germanic, though at a period later than the beginning of the “sound-shifting.” This extinguished the complex system of Indo-European accentuation, which is directly reflected in Sanskrit, and was itself replaced in Latin and Oscan by another system already mentioned, but not in Latin till it had produced marked effects upon the language (e.g. the degradation of the vowels in compounds as in cōnficio from cón-facio, inclūdo from ín-claudo). This curious wave of accentual change (first pointed out by Dieterich, Kuhn’s Zeitschrift, i., and later by Thurneysen, Revue celtique, vi. 312, Rheinisches Museum, xliii. 349) needs and deserves to be more closely investigated from a chronological standpoint. At present it is not clear how far it was a really connected process in all the languages. (See further Kretschmer, op. cit. p. 115, K. Brugmann, Kurze vergleichende Grammatik (1902-1904), p. 57, and their citations, especially Meyer-Lübke, Die Betonung im Gallischen (1901).)

(i.) Extension of the abstract-noun stems in -ti- (like Greek φάτις with Attic βάσις, &c.) by an -n- suffix, as in Lat. mentio (stem mentiōn-) = Ir. (er-)mitiu (stem miti-n-), contrasted with the same word without the n-suffix in Sans. mati-, Lat. mens, Ind.-Eur. *mṇ-ti-. A similar extension (shared also by Gothic) appears in Lat. iuventū-t-, O. Ir. óitiu (stem oiliūt-) beside the simple -tu- in nouns like senātus.

(ii.) Superlative formation in -is-ṃmo- as in Lat. aegerrimus for *aegr-isṃmos, Gallic Οὐξισάμη the name of a town meaning “the highest.”

(iii.) Genitive singular of the o-stems (second declension) in -ī Lat. agri, O. Ir. (Ogam inscriptions) magi, “of a son.”

(iv.) Passive and deponent formation in -r, Lat. sequitur = Ir. sechedar, “he follows.” The originally active meaning of this curious -r suffix was first pointed out by Zimmer (Kuhn’s Zeitschrift, 1888, xxx. 224), who thus explained the use of the accusative pronouns with these “passive” forms in Celtic; Ir. -m-berar, “I am carried,” literally “folk carry me”; Umb. pir ferar, literally ignem feratur, though as pir is a neuter word (= Gr. πῦρ) this example was not so convincing. But within a twelvemonth of the appearance of Zimmer’s article, an Oscan inscription (Conway, Camb. Philol. Society’s Proceedings, 1890, p. 16, and Italic Dialects, p. 113) was discovered containing the phrase ůltiumam (iůvilam) sakrafỉr, “ultimam (imaginem) consecraverint” (or “ultima consecretur”) which demonstrated the nature of the suffix in Italic also. This originally active meaning of the -r form (in the third person singular passive) is the cause of the remarkable fondness for the “impersonal” use of the passive in Latin (e.g., itur in antiquam silvam, instead of eunt), which was naturally extended to all tenses of the passive (ventum est, &c.), so soon as its origin was forgotten. Fuller details of the development will be found in Conway, op. cit. p. 561, and the authorities there cited (very little is added by K. Brugmann, Kurze vergl. Gramm. 1904, p. 596).

(v.) Formation of the perfect passive from the -to- past participle, Lat. monitus (est), &c., Ir. léic-the, “he was left,” ro-léiced, “he has been left.” In Latin the participle maintains its distinct adjectival character; in Irish (J. Strachan, Old Irish Paradigms, 1905, p. 50) it has sunk into a purely verbal form, just as the perfect participles in -us in Umbrian have been absorbed into the future perfect in -ust (entelust, “intenderit”; benust, “venerit”) with its impersonal passive or third plural active -us(s)so (probably standing for -ussor) as in benuso, “ventum erit” (or “venerint”).

To these must be further added some striking peculiarities in phonology.

(vi.) Assimilation of p to a qṷ in a following syllable as in Lat. quinque = Ir. cóic, compared with Sans. pánca, Gr. πέντε, Eng. five, Ind.-Eur. *penqe.

(vii.) Finally—and perhaps this parallelism is the most important of all from the historical standpoint—both Italic and Celtic are divided into two sub-families which differ, and differ in the same way, in their treatment of the Ind.-Eur. velar tenuis q. In both halves of each group it was labialized to some extent; in one half of each group it was labialized so far as to become p. This is the great line of cleavage (i.) between Latinian (Lat. quod, quandō, quinque; Falisc. cuando) and Osco-Umbrian, better called Safine (Osc. pod, Umb. panū- [for *pandō], Osc.-Umb. pompe-, “five,” in Osc. půmperias “nonae,” Umb. pumpeḓia-, “fifth day of the month”); and (ii.) between Goidelic (Gaelic) (O. Ir. cóic, “five,” maq, “son”; modern Irish and Scotch Mac as in MacPherson) and Brythonic (Britannic) (Welsh pump, “five,” Ap for map, as in Powel for Ap Howel).

The same distinction appears elsewhere; Germanic belongs, broadly described, to the q-group, and Greek, broadly described, to the p-group. The ethnological bearing of the distinction within Italy is considered in the articles Sabini and Volsci; but the wider questions which the facts suggest have as yet been only scantily discussed; see the references for the “Sequanian” dialect of Gallic (in the inscription of Coligny, whose language preserves q) in the article Celts: Language.

From these primitive affinities we must clearly distinguish the numerous words taken into Latin from the Celts of north Italy within the historic period; for these see especially an interesting study by J. Zwicker, De vocabulis et rebus Gallicis sive Transpadanis apud Vergilium (Leipzig dissertation, 1905).

I. Phonology

10. The Latin Accent.—It will be convenient to begin with some account of the most important discovery made since the application of scientific method to the study of Latin, for, though it is not strictly a part of phonology, it is wrapped up with much of the development both of the sounds and, by consequence, of the inflexions. It has long been observed (as we have seen § 4, iv. above) that the restriction of the word-accent in Latin to the last three syllables of the word, and its attachment to a long syllable in the penult, were certainly not its earliest traceable condition; between this, the classical system, and the comparative freedom with which the word-accent was placed in pro-ethnic Indo-European, there had intervened a period of first-syllable accentuation to which were due many of the characteristic contractions of Oscan and Umbrian, and in Latin the degradation of the vowels in such forms as accentus from ad + cantus or praecipitem from prae + caput- (§ 19 below). R. von Planta (Osk.-Umbr. Grammatik, 1893, i. p. 594) pointed out that in Oscan also, by the 3rd century B.C., this first-syllable-accent had probably given way to a system which limited the word-accent in some such way as in classical Latin. But it remained for C. Exon, in a brilliant article (Hermathena (1906), xiv. 117, seq.), to deduce from the more precise stages of the change (which had been gradually noted, see e.g. F. Skutsch in Kroll’s Altertumswissenschaft im letzten Vierteljahrhundert, 1905) their actual effect on the language.

11. Accent in Time of Plautus.—The rules which have been established for the position of the accent in the time of Plautus are these:

Exon’s Laws of Syncope.—With these facts are now linked what may be called Exon’s Laws, viz:—

In pre-Plautine Latin in all words or word-groups of four or more syllables whose chief accent is on one long syllable, a short unaccented medial vowel was syncopated; thus *quínquedecem became *quínqdecem and thence quíndecim (for the -im see § 19), *súps-emere became *súpsmere and that sūmere (on -psm- v. inf.) *súrregere, *surregémus, and the like became surgere, surgémus, and the rest of the paradigm followed; so probably validé bonus became valdé bonus, exterá viam became extrá viam; so *supo-téndo became subtendo (pronounced sup-tendo), *āridére, *avidére (from āridus, avidus) became ārdére, audére. But the influence of cognate forms often interfered; posterí-diē became postrídiē, but in posterórum, posterárum the short syllable was restored by the influence of the trisyllabic cases, pósterus, pósterī, &c., to which the law did not apply. Conversely, the nom. *áridor (more correctly at this period *āridōs), which would not have been contracted, followed the form of ārdórem (from *āridórem), ārdére, &c.

The same change produced the monosyllabic forms nec, ac, neu, seu, from neque, &c., before consonants, since they had no accent of their own, but were always pronounced in one breath with the following word, neque tántum becoming nec tantum, and the like. So in Plautus (and probably always in spoken Latin) the words nemp(e), ind(e), quipp(e), ill(e), are regularly monosyllables.

12. Syncope of Final Syllables.—It is possible that the frequent but far from universal syncope of final syllables in Latin (especially before -s, as in mēns, which represents both Gr. μένος and Sans. matís = Ind.-Eur. mṇtís, Eng. mind) is due also to this law operating on such combinations as bona mēns and the like, but this has not yet been clearly shown. In any case the effects of any such phonetic change have been very greatly modified by analogical changes. The Oscan and Umbrian syncope of short vowels before final s seems to be an independent change, at all events in its detailed working. The outbreak of the unconscious affection of slurring final syllables may have been contemporaneous.

13. In post-Plautine Latin words accented on the ante-antepenult:—

(i.) suffered syncope in the short syllable following the accented syllable (bálineae became bálneae, puéritia became puértia (Horace), cólumine, tégimine, &c., became cúlmine, tégmine, &c., beside the trisyllabic cólumen, tégimen) unless

(ii.) that short vowel was e or i, followed by another vowel (as in párietem, múlierem, Púteoli), when, instead of contraction, the accent shifted to the penult, which at a later stage of the language became lengthened, pariétem giving Ital. paréếte, Fr. paroi, Puteóli giving Ital. Pozzuốli.

The restriction of the accent to the last three syllables was completed by these changes, which did away with all the cases in which it had stood on the fourth syllable.

14. The Law of the Brevis Brevians.—Next must be mentioned another great phonetic change, also dependent upon accent, which had come about before the time of Plautus, the law long known to students as the Brevis Brevians, which may be stated as follows (Exon, Hermathena (1903), xii. 491, following Skutsch in, e.g., Vollmöller’s Jahresbericht für romanische Sprachwissenschaft, i. 33): a syllable long by nature or position, and preceded by a short syllable, was itself shortened if the word-accent fell immediately before or immediately after it—that is, on the preceding short syllable or on the next following syllable. The sequence of syllables need not be in the same word, but must be as closely connected in utterance as if it were. Thus mốdō became módŏ, vŏlūptấtēm became vŏlŭ(p)tấtem, quḯd ēst? became quid ĕst? either the s or the t or both being but faintly pronounced.

It is clear that a great number of flexional syllables so shortened would have their quantity immediately restored by the analogy of the same inflexion occurring in words not of this particular shape; thus, for instance, the long vowel of ấmā and the like is due to that in other verbs (pulsā, agitā) not of iambic shape. So ablatives like modö, sonō get back their -ō, while in particles like modo, “only,” quōmodo, “how,” the shortened form remains. Conversely, the shortening of the final -a in the nom. sing. fem. of the a-declension (contrast lūnă with Gr. χωρᾷ) was probably partly due to the influence of common forms like eă, bonă, mală, which had come under the law.

15. Effect on Verb Inflexion.—These processes had far-reaching effects on Latin inflexion. The chief of these was the creation of the type of conjugation known as the capio-class. All these verbs were originally inflected like audio, but the accident of their short root-syllable, (in such early forms as *fúgīs, *fugītṹrus, *fugīsếtis, &c., becoming later fúgĭs, fugĭtṹrus, fugĕrếtis) brought great parts of their paradigm under this law, and the rest followed suit; but true forms like fugīre, cupīre, morīri, never altogether died out of the spoken language. St Augustine, for instance, confessed in 387 A.D. (Epist. iii. 5, quoted by Exon, Hermathena (1901), xi. 383,) that he does not know whether cupi or cupiri is the pass. inf. of cupio. Hence we have Ital. fuggīre, morīre, Fr. fuir, mourir. (See further on this conjugation, C. Exon, l.c., and F. Skutsch, Archiv für lat. Lexicographie, xii. 210, two papers which were written independently.)

16. The question has been raised how far the true phonetic shortening appears in Plautus, produced not by word-accent but by metrical ictus—e.g. whether the reading is to be trusted in such lines as Amph. 761, which gives us dedisse as the first foot (tribrach) of a trochaic line “because the metrical ictus fell on the syllable ded-”—but this remarkable theory cannot be discussed here. See the articles cited and also F. Skutsch, Forschungen zu Latein. Grammatik und Metrik, i. (1892); C. Exon, Hermathena (1903) xii. p. 492, W. M. Lindsay, Captivi (1900), appendix.

In the history of the vowels and diphthongs in Latin we must distinguish the changes which came about independently of accent and those produced by the preponderance of accent in another syllable.

17. Vowel Changes independent of Accent.—In the former category the following are those of chief importance:—

(i.) ĭ became ĕ (a) when final, as in ant-e beside Gr. ἀντί, trīste besides trīsti-s, contrasted with e.g., the Greek neuter ἴδρι (the final -e of the infinitive—regere, &c.—is the -ĭ of the locative, just as in the so-called ablatives genere, &c.); (b) before -r- which has arisen from -s-, as in cineris beside cinis, cinisculus; serō beside Gr. ἴ(σ)ημι (Ind.-Eur. *si-sēmi, a reduplicated non-thematic present).

(ii.) Final ŏ became ĕ; imperative sequere = Gr. ἔπε(σ)ο; Lat. ille may contain the old pronoun *so, “he,” Gr. ὁ, Sans. sa (otherwise Skutsch, Glotta, i. Hefte 2-3).

(iii.) el became ol when followed by any sound save e, i or l, as in volō, volt beside velle; colō beside Gr. τέλλομαι, πολεῖν, Att. τέλος; colōnus for *quelōnus, beside inquilīnus for *en-quēlenus.

(iv.) e became i (i.) before a nasal followed by a palatal or velar consonant (tingo, Gr. τέγγω; in-cipio from *en-capio); (ii.) under certain conditions not yet precisely defined, one of which was i in a following syllable (nihil, nisi, initium). From these forms in- spread and banished en-, the earlier form.

(v.) The “neutral vowel” (“schwa Indo-Germanicum”) which arose in pro-ethnic Indo-European from the reduction of long ā, ē or ō in unaccented syllables (as in the -tós participles of such roots as stā-, dhē-, dō-, *stƏtós, *dhƏtós, *dƏtós) became a in Latin (status con-ditus [from *con-dhatos], datus), and it is the same sound which is represented by a in most of the forms of dō (damus, dabō, &c.).

(vi.) When a long vowel came to stand before another vowel in the same word through loss of ḭ or ṷ, it was always shortened; thus the -eō of intransitive verbs like candeō, caleō is for -ēḭō (where the ē is identical with the η in Gr. ἐφάνην, ἐμάνμν) and was thus confused with the causative -eiō (as in moneō, “I make to think,” &c.), where the short e is original. So audīuī became audīī and thence audiī (the form audīvī would have disappeared altogether but for being restored from audīveram, &c.; conversely audieram is formed from audiī). In certain cases the vowels contracted, as in trēs, partēs, &c. with -ēs from eḭes, *amō from amā(ḭ)ō.

18. Of the Diphthongs.

(vii.) eu became ou in pro-ethnic Italic, Lat. novus: Gr. νέος, Lat. novem, Umb. nuviper (i.e. noviper), “usque ad Changes of the diphthongs independent of accent. noviens”: Gr. (ἐν-)νέα; in unaccented syllables this -ov- sank to -u(v)- as in dếnuō from dế novō, suus (which is rarely anything but an enclitic word), Old Lat. sovos: Gr. ἑ(ϝ)ός.

(viii.) ou, whether original or from eu, when in one syllable became -ū-, probably about 200 B.C., as in dūcō, Old Lat. doucō, Goth, tiuhan, Eng. tow, Ind.-Eur. *deṷcō.

(ix.) ei became ī (as in dīcō, Old Lat. deico: Gr. δείκ-νυμι, fīdo: Gr. πείθομαι, Ind.-Eur. *bheidhō) just before the time of Lucilius, who prescribes the spellings puerei (nom. plur.) but puerī (gen. sing.), which indicates that the two forms were pronounced alike in his time, but that the traditional distinction in spelling had been more or less preserved. But after his time, since the sound of ei was merely that of ī, ei is continually used merely to denote a long ī, even where, as in faxeis for faxīs, there never had been any diphthongal sound at all.

(x.) In rustic Latin (Volscian and Sabine) au became ō as in the vulgar terms explōdere, plōstrum. Hence arose interesting doublets of meaning;—lautus (the Roman form), “elegant,” but lōtus, “washed”; haustus, “draught,” but hōstus (Cato), “the season’s yield of fruit.”

(xi.) oi became oe and thence ū some time after Plautus, as in ūnus, Old Lat. oenus: Gr. οἰνή “ace.” In Plautus the forms have nearly all been modernized, save in special cases, e.g. in Trin. i. 1, 2, immoene facinus, “a thankless task,” has not been changed to immune because that meaning had died out of the adjective so that immune facinus would have made nonsense; but at the end of the same line utile has replaced oetile. Similarly in a small group of words the old form was preserved through their frequent use in legal or religious documents where tradition was strictly preserved—poena, foedus (neut.), foedus (adj.), “ill-omened.” So the archaic and poetical moenia, “ramparts,” beside the true classical form mūnia, “duties”; the historic Poeni beside the living and frequently used Pūnicum (bellum)—an example which demonstrates conclusively (pace Sommer) that the variation between ū and oe is not due to any difference in the surrounding sounds.

(xii.) ai became ae and this in rustic and later Latin (2nd or 3rd century A.D.) simple ē, though of an open quality—Gr. αἴθος, αἴθω, Lat. aedēs (originally “the place for the fire”); the country forms of haedus, praetor were edus, pretor (Varro, Ling. Lat. v. 97, Lindsay, Lat. Lang. p. 44).

19. Vowels and Diphthongs in unaccented Syllables.—The changes of the short vowels and of the diphthongs in unaccented syllables are too numerous and complex to be set forth here. Some took place under the first-syllable system of accent, some later (§§ 9, 10). Typical examples are pepErci from *péparcai and ónustus from *ónostos (before two consonants); concIno from *cóncano and hospItIs from *hóstipotes, legImus beside Gr. λέγομεν (before one consonant); SicUli from *Siceloi (before a thick l, see § 17, 3); dilIgIt from *dísleget (contrast, however, the preservation of the second e in neglEgIt); occUpat from *opcapat (contrast accipit with i in the following syllable); the varying spelling in monumentum and monimentum, maxumus and maximus, points to an intermediate sound (ü) between u and i (cf. Quint. i. 4. 8, reading optumum and optimum [not opimum] with W. M. Lindsay, Latin Language §§ 14, 16, seq.), which could not be correctly represented in spelling; this difference may, however, be due merely to the effect of differences in the neighbouring sounds, an effect greatly obscured by analogical influences.

Inscriptions of the 4th or 3rd century, B.C. which show original -es and -os in final syllables (e.g. Venerĕs, gen. sing., nāvebos abl. pl.) compared with the usual forms in -is, -us a century later, give us roughly the date of these changes. But final -os, -om, remained after -u- (and v) down to 50 B.C. as in servos.

20. Special mention should be made of the change of -rĭ- and -ro- to -er- (incertus from *encritos; ager, ācer from *agros, *ācris; the feminine ācris was restored in Latin (though not in North Oscan) by the analogy of other adjectives, like tristis, while the masculine ācer was protected by the parallel masculine forms of the -o- declension, like tener, niger [from *teneros, *nigros]).

21. Long vowels generally remained unchanged, as in compāgo, condōno.

22. Of the diphthongs, ai and oi both sank to ei, and with original ei further to ī, in unaccented syllables, as in Achivi from Gr. Ἀχαιϝοί, olīivom, earlier *oleivom (borrowed into Gothic and there becoming alēv) from Gr. ἔλαιϝον. This gives us interesting chronological data, since the el- must have changed to ol- (§ 16. 3) before the change of -ai- to -ei-, and that before the change of the accent from the first syllable to the penultimate (§ 9); and the borrowing took place after -ai- had become -ei-, but before -eivom had become -eum, as it regularly did before the time of Plautus.

But cases of ai, ae, which arose later than the change to ei, ī, were unaffected by it; thus the nom. plur. of the first declension originally ended in -ās (as in Oscan), but was changed at some period before Plautus to -ae by the influence of the pronominal nom. plur. ending -ae in quae? hae, &c., which was accented in these monosyllables and had therefore been preserved. The history of the -ae of the dative, genitive and locative is hardly yet clear (see Exon, Hermathena (1905), xiii. 555; K. Brugmann, Grundriss, 1st ed. ii. 571, 601).

The diphthongs au, ou in unaccented syllables sank to -u-, as in inclūdō beside claudō; the form clūdō, taken from the compounds, superseded claudo altogether after Cicero’s time. So cūdō, taken from incūdō, excūdō, banished the older *caudō, “I cut, strike,” with which is probably connected cauda, “the striking member, tail,” and from which comes caussa, “a cutting, decision, legal case,” whose -ss- shows that it is derived from a root ending in a dental (see §25 (b) below and Conway, Verner’s Law in Italy, p. 72).

Consonants.—Passing now to the chief changes of the consonants we may notice the following points:—

23. Consonant i (wrongly written j; there is no g-sound in the letter), conveniently written ḭ by phoneticians,

(i.) was lost between vowels, as in trēs for *treḭes, &c. (§ 17. 6);

(ii.) in combination: -mḭ- became -ni-, as in veniö, from Ind.-Eur. *Ƨṷ mḭo, “I come,” Sans. gam-, Eng. come; -nḭ- probably (under certain conditions at least) became -nd-, as in tendō beside Gr. τείνω, fendō = Gr. θείνω, and in the gerundive stem -endus, -undus, probably for -enḭos, -onḭos; cf. the Sanskrit gerundive in -an-īya-s; -gḭ-, -dḭ- became -ḭ- as in māior from *mag-ior, pēior from *ped-ior;

(iii.) otherwise -ḭ- after a consonant became generally syllabic (-iḭ-), as in capiō (trisyllabic) beside Goth. hafya.

24. Consonant u (formerly represented by English v), conveniently written ṷ,

(i.) was lost between similar vowels when the first was accented, as in audīui, which became audiī (§ 17 [6]), but not in amāuī, nor in avārus.

(ii.) in combination: dṷ became b, as in bonus, bellum, O. Lat. dṷonus, *dṷellum (though the poets finding this written form in old literary sources treated it as trisyllabic); pṷ-, fṷ-, bṷ-, lost the ṷ, as in ap-erio, op-erio beside Lith. -veriu, “I open,” Osc. veru, “gate,” and in the verbal endings -bam, -bō, from -bhṷ-ām, -bhṷō (with the root of Lat. fui), and fīo, du-bius, super-bus, vasta-bundus, &c., from the same; -sṷ- between vowels (at least when the second was accented) disappeared (see below § 25 (a), iv.), as in pruīna for prusuīna, cf. Eng. fros-t, Sans, pruṣvā, “hoar-frost.” Contrast Minérva from an earlier *menes-ṷā, sṷe-, sṷo-, both became so-, as in sorōor(em) beside Sans. svasār-am, Ger. schwes-t-er, Eng. sister, sordēs, beside O. Ger. swart-s, mod. schwarz. -ṷo- in final syllables became -u-, as in cum from quom, parum from parṷom; but in the declensional forms -ṷu- was commonly restored by the analogy of the other cases, thus (a) serṷos, serṷom, serṷī became (b) *serus, *serum, *serṷi, but finally (c) serṷus, serṷum, serṷi.

(iii.) In the 2nd century A.D., Lat. v (i.e. ṷ) had become a voiced labio-dental fricative, like Eng. v; and the voiced labial plosive b had broken down (at least in certain positions) into the same sound; hence they are frequently confused as in spellings like vene for bene, Bictorinus for Victorinus.

25. (a) Latin s

(i.) became r between vowels between 450 and 350 B.C. (for the date see R. S. Conway, Verner’s Law in Italy, pp. 61-64), as āra, beside O. Lat. āsa, generis from *geneses, Gr. γένεος; eram, erō for *esām, *esō, and so in the verbal endings -erām, -erō, -erim. But a considerable number of words came into Latin, partly from neighbouring dialects, with -s- between vowels, after 350 B.C., when the change ceased, and so show -s-, as rosa (probably from S. Oscan for *rodḭa “rose-bush” cf. Gr. ῥόδον), cāseus, “cheese,” miser, a term of abuse, beside Gr. μυσαρός (probably also borrowed from south Italy), and many more, especially the participles in -sus (fūsus), where the -s- was -ss- at the time of the change of -s- to -r- (so in causa, see above). All attempts to explain the retention of the -s- otherwise must be said to have failed (e.g. the theory of accentual difference in Verner’s Law in Italy, or that of dissimilation, given by Brugmann, Kurze vergl. Gram. p. 242).

(ii.) sr became þr (= Eng. thr in throw) in pro-ethnic Italic, and this became initially fr- as in frīgus, Gr. ῥῖγος (Ind.-Eur. *srīgos), but medially -br-, as in funebris, from funus, stem funes-.

(iii.) -rs-, ls- became -rr-, -ll-, as in ferre, velle, for *fer-se, *vel-se (cf. es-se).

(iv.) Before m, n, l, and v, -s- vanished, having previously caused the loss of any preceding plosive or -n-, and the preceding vowel, if short, was lengthened as in

(v.) Similarly -sd- became -d-, as in īdem from is-dem.

(vi.) Before n-, m-, l-, initially s- disappeared, as in nūbo beside Old Church Slavonic snubiti, “to love, pay court to”; mīror beside Sans, smáyatē, “laughs,” Eng. smi-le; lūbricus beside Goth, sliupan, Eng. slip.

(b) Latin -ss- arose from an original -t + t-, -d + t-, -dh + t- (except before -r), as in missus, earlier *mit-tos; tōnsus, earlier *tond-tos, but tonstrīx from *tond-trīx. After long vowels this -ss- became a single -s- some time before Cicero (who wrote caussa [see above], divissio, &c., but probably only pronounced them with -s-, since the -ss- came to be written single directly after his time).

26. Of the Indo-European velars the breathed q was usually preserved in Latin with a labial addition of -ṷ- (as in sequor, Gr. ἕπομαι, Goth, saihvan, Eng. see; quod, Gr. ποδ-(απός), Eng. what); but the voiced Ƨṷ remained (as -gu-) only after -n- (unguo beside Ir. imb, “butter”) and (as g) before r, l, and u (as in gravis, Gr. βαρύς; glans, Gr. βάλανος; legūmen, Gr. λοβός, λεβίνθος). Elsewhere it became v, as in veniō (see § 23, ii.), nūdus from *novedos, Eng. naked. Hence bōs (Sans. gāus, Eng. cow) must be regarded as a farmer’s word borrowed from one of the country dialects (e.g. Sabine); the pure Latin would be *vōs, and its oblique cases, e.g. acc. *vovem, would be inconveniently close in sound to the word for sheep ovem.

27. The treatment of the Indo-European voiced aspirates (bh, dh, ḡh Ƨh) in Latin is one of the most marked characteristics of the language, which separates it from all the other Italic dialects, since the fricative sounds, which represented the Indo-European aspirates in pro-ethnic Italic, remained fricatives medially if they remained at all in that position in Oscan and Umbrian, whereas in Latin they were nearly always changed into voiced explosives. Thus—

For the “non-labializing velars” (Hostis, conGius, Glaber) reference must be made to the fuller accounts in the handbooks.

28. Authorities.—This summary account of the chief points in Latin phonology may serve as an introduction to its principles, and give some insight into the phonetic character of the language. For systematic study reference must be made to the standard books, Karl Brugmann, Grundriss der vergleichenden Grammatik der Indo-Germanischen Sprachen (vol. i., Lautlehre, 2nd ed. Strassburg, 1897; Eng. trans. of ed. 1 by Joseph Wright, Strassburg, 1888) and his Kurze vergleichende Grammatik (Strassburg, 1902); these contain still by far the best accounts of Latin; Max Niederman, Précis de phonétique du Latin (Paris, 1906), a very convenient handbook, excellently planned; F. Sommer, Lateinische Laut- und Flexionslehre (Heidelberg, 1902), containing many new conjectures; W. M. Lindsay, The Latin Language (Oxford, 1894), translated into German (with corrections) by Nohl (Leipzig, 1897), a most valuable collection of material, especially from the ancient grammarians, but not always accurate in phonology; F. Stolz, vol. i. of a joint Historische Grammatik d. lat. Sprache by Blase, Landgraf, Stolz and others (Leipzig, 1894); Neue-Wagener, Formenlehre d. lat. Sprache (3 vols., 3rd ed. Leipzig, 1888, foll.); H. J. Roby’s Latin Grammar (from Plautus to Suetonius; London, 7th ed., 1896) contains a masterly collection of material, especially in morphology, which is still of great value. W. G. Hale and C. D. Buck’s Latin Grammar (Boston, 1903), though on a smaller scale, is of very great importance, as it contains the fruit of much independent research on the part of both authors; in the difficult questions of orthography it was, as late as 1907, the only safe guide.

II. Morphology

In morphology the following are the most characteristic Latin innovations:—

29. In nouns.

(i.) The complete loss of the dual number, save for a survival in the dialect of Praeneste (C.I.L. xiv. 2891, = Conway, Ital. Dial. p. 285, where Q. k. Cestio Q. f. seems to be nom. dual); so C.I.L. xi. 67065, T. C. Vomanio, see W. Schulze, Lat. Eigennamen, p. 117.

(ii.) The introduction of new forms in the gen. sing, of the -o- stems (dominī), of the -ā- stems (mēnsae) and in the nom. plural of the same two declensions; innovations mostly derived from the pronominal declension.

(iii.) The development of an adverbial formation out of what was either an instrumental or a locative of the -o- stems, as in longē. And here may be added the other adverbial developments, in -m (palam, sensim) probably accusative, and -iter, which is simply the accusative of iter, “way,” crystallized, as is shown especially by the fact that though in the end it attached itself particularly to adjectives of the third declension (molliter), it appears also from adjectives of the second declension whose meaning made their combination with iter especially natural, such as longiter, firmiter, largiter (cf. English straightway, longways). The only objections to this derivation which had any real weight (see F. Skutsch, De nominibus no- suffixi ope formatis, 1890, pp. 4-7) have been removed by Exon’s Law (§ 11), which supplies a clear reason why the contracted type constanter arose in and was felt to be proper to Participial adverbs, while firmiter and the like set the type for those formed from adjectives.

(iv.) The development of the so-called fifth declension by a re-adjustment of the declension of the nouns formed with the suffix -iē-: ia- (which appears, for instance, in all the Greek feminine participles, and in a more abstract sense in words like māteriēs) to match the inflexion of two old root-nouns rēs and diēs, the stems of which were originally rēḭ- (Sans. rās, rāyas, cf. Lat. reor) and diēṷ-.

(v.) The disuse of the -ti- suffix in an abstract sense. The great number of nouns which Latin inherited formed with this suffix were either (1) marked as abstract by the addition of the further suffix -ōn- (as in natio beside the Gr. γνὴσι-ος, &c.) or else (2) confined to a concrete sense; thus vectis, properly “a carrying, lifting,” came to mean “pole, lever”; ratis, properly a “reckoning, devising,” came to mean “an (improvised) raft” (contrast ratiō); postis, a “placing,” came to mean “post.”

(vi.) The confusion of the consonantal stems with stems ending in -ĭ-. This was probably due very largely to the forms assumed through phonetic changes by the gen. sing. and the nom. and acc. plural. Thus at say 300 B.C. the inflexions probably were:

	conson. stem	-ĭ- stem
Nom. plur.	*rēg-ĕs	host-ēs
Acc. plur.	rĕg-ēs	host-īs

The confusing difference of signification of the long -ēs ending led to a levelling of these and other forms in the two paradigms.

(vii.) The disuse of the u declension (Gr. ἡδύς, στάχυς) in adjectives; this group in Latin, thanks to its feminine form (Sans. fem. svādvī, “sweet”), was transferred to the i declension (suavis, gravis, levis, dulcis).

30. In verbs.

(i.) The disuse of the distinction between the personal endings of primary and secondary tenses, the -t and -nt, for instance, being used for the third person singular and plural respectively in all tenses and moods of the active. This change was completed after the archaic period, since we find in the oldest inscriptions -d regularly used in the third person singular of past tenses, e.g. deded, feced in place of the later dedit, fecit; and since in Oscan the distinction was preserved to the end, both in singular and plural, e.g. faamat (perhaps meaning “auctionatur”), but deded (“dedit”). It is commonly assumed from the evidence of Greek and Sanskrit (Gr. ἕστι, Sans. asti beside Lat. est) that the primary endings in Latin have lost a final -i, partly or wholly by some phonetic change.

(ii.) The non-thematic conjugation is almost wholly lost, surviving only in a few forms of very common use, est, “is”; ēst, “eats”; volt, “wills,” &c.

(iii.) The complete fusion of the aorist and perfect forms, and in the same tense the fusion of active and middle endings; thus tutudī, earlier *tutudai, is a true middle perfect; dīxī is an s aorist with the same ending attached; dīxit is an aorist active; tutudisti is a conflation of perfect and aorist with a middle personal ending.

(iv.) The development of perfects in -uī and -vī, derived partly from true perfects of roots ending in v or u, e.g. mōvī ruī. For the origin of monuī see Exon, Hermathena (1901), xi. 396 sq.

(v.) The complete fusion of conjunctive and optative into a single mood, the subjunctive; regam, &c., are conjunctive forms, whereas rexerim, rexissem are certainly and regerem most probably optative; the origin of amem and the like is still doubtful. Notice, however, that true conjunctive forms were often used as futures, regēs, reget, &c., and also the simple thematic conjunctive in forms like erō, rexerō, &c.

(vi.) The development of the future in -bo and imperfect in -bam by compounding some form of the verb, possibly the Present Participle with forms from the root of fuī, *amans-fuo becoming amabō, *amans-fṷām becoming amābam at a very early period of Latin; see F. Skutsch, Atti d. Congresso Storico Intern. (1903), vol. ii. p. 191.

(vii.) We have already noticed the rise of the passive in -r (§ 5 (d)). Observe, however, that several middle forms have been pressed into the service, partly because the -r- in them which had come from -s- seemed to give them a passive colour (legere = Gr. λέγε(σ)ο, Attic λέγου). The interesting forms in -minī are a confusion of two distinct inflexions, namely, an old infinitive in -menai, used for the imperative, and the participial -menoi, masculine, -menai, feminine, used with the verb “to be” in place of the ordinary inflexions. Since these forms had all come to have the same shape, through phonetic change, their meanings were fused; the imperative forms being restricted to the plural, and the participial forms being restricted to the second person.

31. Past Participle Passive.—Next should be mentioned the great development in the use of the participle in -tos (factus, fusus, &c.). This participle was taken with sum to form the perfect tenses of the passive, in which, thanks partly to the fusion of perfect and aorist active, a past aorist sense was also evolved. This reacted on the participle itself giving it a prevailingly past colour, but its originally timeless use survives in many places, e.g. in the participle ratus, which has as a rule no past sense, and more definitely still in such passages as Vergil, Georg. i. 206 (vectis), Aen. vi. 22 (ductis), both of which passages demand a present sense. It is to be noticed also that in the earliest Latin, as in Greek and Sanskrit, the passive meaning, though the commonest, is not universal. Many traces of this survive in classical Latin, of which the chief are

32. Future Participle.—Finally may be mentioned together (a) the development of the future participle active (in -ūrus, never so freely used as the other participles, being rare in the ablative absolute even in Tacitus) from an old infinitive in -ūrum (“scio inimicos meos hoc dicturum,” C. Gracchus (and others) apud Gell. 1. 7, and Priscian ix. 864 (p. 475 Keil), which arose from combining the dative or locative of the verbal noun in -tu with an old infinitive esom “esse” which survives in Oscan, *dictu esom becoming dicturum. This was discovered by J. P. Postgate (Class. Review, v. 301, and Idg. Forschungen iv. 252). (b) From the same infinitival accusative with the post-position -dō, meaning “to,” “for,” “in” (cf. quandō for *quam-do, and Eng. to, Germ, zu) was formed the so-called gerund agen-dō, “for doing,” “in doing,” which was taken for a Case, and so gave rise to the accusative and genitive in -dum and -dī. The form in -dō still lives in Italian as an indeclinable present participle. The modal and purposive meanings of -dō appear in the uses of the gerund.

The authorities giving a fuller account of Latin morphology are the same as those cited in § 28 above, save that the reader must consult the second volume of Brugmann’s Grundriss, which in the English translation (by Conway and Rouse, Strassburg, 1890-1896) is divided into volumes ii, iii. and iv.; and that Niedermann does not deal with morphology.

III. Syntax

The chief innovations of syntax developed in Latin may now be briefly noted.

33. In nouns.

(i.) Latin restricted the various Cases to more sharply defined uses than either Greek or Sanskrit; the free use of the internal accusative in Greek (e.g. ἁβρὸν βαίνειν, τυφλὸς τὰ ὦτα) is strange to Latin, save in poetical imitations of Greek; and so is the freedom of the Sanskrit instrumental, which often covers meanings expressed in Latin by cum, ab, inter.

(ii.) The syncretism of the so-called ablative case, which combines the uses of (a) the true ablative which ended in -d (O. Lat. praidād); (b) the instrumental sociative (plural forms like dominīs, the ending being that of Sans. çivāiş); and (c) the locative (noct-e, “at night”; itiner-e, “on the road,” with the ending of Greek ἐλπίδ-ι). The so-called absolute construction is mainly derived from the second of these, since it is regularly attached fairly closely to the subject of the clause in which it stands, and when accompanied by a passive participle most commonly denotes an action performed by that subject. But the other two sources cannot be altogether excluded (orto sole, “starting from sunrise”; campo patente, “on, in sight of, the open plain”).

34. In verbs.

(i.) The rich development and fine discrimination of the uses of the subjunctive mood, especially (a) in indirect questions (based on direct deliberative questions and not fully developed by the time of Plautus, who constantly writes such phrases as dic quis es for the Ciceronian dic quis sis); (b) after the relative of essential definition (non is sum qui negem) and the circumstantial cum (“at such a time as that”). The two uses (a) and (b) with (c) the common Purpose and Consequence-clauses spring from the “prospective” or “anticipatory” meaning of the mood. (d) Observe further its use in subordinate oblique clauses (irascitur quod abierim, “he is angry because, as he asserts, I went away”). This and all the uses of the mood in oratio obliqua are derived partly from (a) and (b) and partly from the (e) Unreal Jussive of past time (Non illi argentum redderem? Non redderes, “Ought I not to have returned the money to him?” “You certainly ought not to have,” or, more literally, “You were not to”).

On this interesting chapter of Latin syntax see W. G. Hale’s “Cum-constructions” (Cornell University Studies in Classical Philology, No. 1, 1887-1889), and The Anticipatory Subjunctive (Chicago, 1894).

(ii.) The complex system of oratio obliqua with the sequence of tenses (on the growth of the latter see Conway, Livy II., Appendix ii., Cambridge, 1901).

(iii.) The curious construction of the gerundive (ad capiendam urbem), originally a present (and future?) passive participle, but restricted in its use by being linked with the so-called gerund (see § 32, b). The use, but probably not the restriction, appears in Oscan and Umbrian.

(iv.) The favourite use of the impersonal passive has already been mentioned (§ 5, iv.).

35. The chief authorities for the study of Latin syntax are: Brugmann’s Kurze vergl. Grammatik, vol. ii. (see § 28); Landgraf’s Historische lat. Syntax (vol. ii. of the joint Hist. Gram., see § 28); Hale and Buck’s Latin Grammar (see § 28); Draeger’s Historische lat. Syntax, 2 vols. (2nd ed., Leipzig, 1878-1881), useful but not always trustworthy; the Latin sections in Delbrück’s Vergleichende Syntax, being the third volume of Brugmann’s Grundriss (§ 28).

(a) The first represents the period of the early intercourse of Rome with the Greek states, especially with the colonies in the south of Italy and Sicily. To this stage belong many names of nations, countries and towns, as Siculi, Tarentum, Graeci, Achivi, Poenus; and also names of weights and measures, articles of industry and terms connected with navigation, as mina, talentum, purpura, patina, ancora, aplustre, nausea. Words like amurca, scutula, pessulus, balineum, tarpessita represent familiarity with Greek customs and bear equally the mark of naturalization. To these may be added names of gods or heroes, like Apollo, Pollux and perhaps Hercules. These all became naturalized Latin words and were modified by the phonetic changes which took place in the Latin language after they had come into it (cf. §§ 9-27 supra). (b) The second stage was probably the result of the closer intercourse resulting from the conquest of southern Italy, and the wars in Sicily, and of the contemporary introduction of imitations of Greek literature into Rome, with its numerous references to Greek life and culture. It is marked by the free use of hybrid forms, whether made by the addition of Latin suffixes to Greek stems as ballistārius, hēpatārius, subbasilicānus, sycophantiōsus, cōmissārī or of Greek suffixes to Latin stems as plāgipatidas, pernōnides; or by derivation, as thermopōtāre, supparasītāri; or by composition as ineuschēmē, thyrsigerae, flagritribae, scrophipascī. The character of many of these words shows that the comic poets who coined them must have been able to calculate upon a fair knowledge of colloquial Greek on the part of a considerable portion of their audience. The most remarkable instance of this is supplied by the burlesque lines in Plautus (Pers. 702 seq.), where Sagaristio describes himself as

Vaniloquidorus, Virginisvendonides, Nugipiloquides, Argentumexterebronides, Tedigniloquides, Nummosexpalponides, Quodsemelarripides, Nunquameripides.

During this period Greek words are still generally inflected according to the Latin usage.

(c) But with Accius (see below) begins a third stage, in which the Greek inflexion is frequently preserved, e.g. Hectora, Oresten, Cithaeron; and from this time forward the practice wavers. Cicero generally prefers the Latin case-endings, defending, e.g., Piraeeum as against Piraeea (ad Att. vii. 3, 7), but not without some fluctuation, while Varro takes the opposite side, and prefers poëmasin to the Ciceronian poëmatis. By this time also y and z were introduced, and the representation of the Greek aspirates by th, ph, ch, so that words newly borrowed from the Greek could be more faithfully reproduced. This is equally true whatever was the precise nature of the sound which at that period the Greek aspirates had reached in their secular process of change from pure aspirates (as in Eng. ant-hill, &c.) to fricatives (like Eng. th in thin). (See Arnold and Conway, The Restored Pronunciation of Greek and Latin, 4th ed., Cambridge, 1908, p. 21.)

(d) A fourth stage is marked by the practice of the Augustan poets, who, especially when writing in imitation of Greek originals, freely use the Greek inflexions, such as Arcaděs, Tethŷ, Aegida, Echūs, &c. Horace probably always used the Latin form in his Satires and Epistles, the Greek in his Odes. Later prose writers for the most part followed the example of his Odes. It must be added, however, in regard to these literary borrowings that it is not quite clear whether in this fourth class, and even in the unmodified forms in the preceding class, the words had really any living use in spoken Latin.

V. Pronunciation

This appears the proper place for a rapid survey of the pronunciation1 of the Latin language, as spoken in its best days.

37. Consonants.—(i.) Back palatal. Breathed plosive c, pronounced always as k (except that in some early inscriptions—probably none much later, if at all later, than 300 B.C.—the character is used also for g) until about the 7th century after Christ. K went out of use at an early period, except in a few old abbreviations for words in which it had stood before a, e.g., kal. for kalendae. Q, always followed by the consonantal u, except in a few old inscriptions, in which it is used for c before the vowel u, e.g. pequnia. X, an abbreviation for cs; xs is, however, sometimes found. Voiced plosive g, pronounced as in English gone, but never as in English gem before about the 6th century after Christ. Aspirate h, the rough breathing as in English.

(ii.) Palatal.—The consonantal i, like the English y; it is only in late inscriptions that we find, in spellings like Zanuario, Giove, any definite indication of a pronunciation like the English j. The precise date of the change is difficult to determine (see Lindsay’s Latin Lang. p. 49), especially as we may, in isolated cases, have before us merely a dialectic variation; see Paeligni.

(iii.) Lingual.—r as in English, but probably produced more with the point of the tongue. l similarly more dental than in English. s always breathed (as Eng. ce in ice). z, which is only found in the transcription of Greek words in and after the time of Cicero, as dz or zz.

(iv.) Dental.—Breathed, t as in English. Voiced, d as in English; but by the end of the 4th century di before a vowel was pronounced like our j (cf. diurnal and journal). Nasal, n as in English; but also (like the English n) a guttural nasal (ng) before a guttural. Apparently it was very lightly pronounced, and easily fell away before s.

(v.) Labial.—Breathed, p as in English. Voiced, b as in English; but occasionally in inscriptions of the later empire v is written for b, showing that in some cases b had already acquired the fricative sound of the contemporary β (see § 24, iii.). b before a sharp s was pronounced p, e.g. in urbs. Nasal, m as in English, but very slightly pronounced at the end of a word. Spirant, v like the ou in French oui, but later approximating to the w heard in some parts of Germany, Ed. Sievers, Grundzüge d. Phonetik, ed. 4, p. 117, i.e. a labial v, not (like the English v) a labio-dental v.

(vi.) Labio-dental.—Breathed fricative, f as in English.

38. Vowels.—ā, ū, ī, as the English ah, oo, ee; ō, a sound coming nearer to Eng. aw than to Eng. ō; ē a close Italian ē, nearly as the a of Eng. mate, ée of Fr. passée. The short sound of the vowels was not always identical in quality with the long sound. ă was pronounced as in the French chatte, ŭ nearly as in Eng. pull, ĭ nearly as in pit, ŏ as in dot, ĕ nearly as in pet. The diphthongs were produced by pronouncing in rapid succession the vowels of which they were composed, according to the above scheme. This gives, au somewhat broader than ou in house; eu like ow in the “Yankee” pronunciation of town; ae like the vowel in hat lengthened, with perhaps somewhat more approximation to the i in wine; oe, a diphthongal sound approximating to Eng. oi; ui, as the French oui.

To this it should be added that the Classical Association, acting on the advice of a committee of Latin scholars, has recommended for the diphthongs ae and oe the pronunciation of English i (really ai) in wine and oi in boil, sounds which they undoubtedly had in the time of Plautus and probably much later, and which for practical use in teaching have been proved far the best.

“iovesat deivos qoi med mitat, nei ted endo cosmis virco sied, asted noisi opetoitesiai pacari vois.

dvenos med feced en manom einom duenoi ne med malo statod.”

“Deos iurat ille (or iurant illi) qui me mittat (or mittant) ne in te Virgo (i.e. Proserpina) comis sit, nisi quidem optimo (?) Theseae (?) pacari vis. Duenos me fecit contra Manum, Dueno autem ne per me malum stato (= imputetur, imponatur).”

R. S. Conway examined it in situ in company with F. Skutsch in 1903 (cf. his article in Vollmöller’s Jahresbericht, vi. 453), and the only words that can be regarded as reasonably certain are regei (regi) on face 2, kalatorem and iouxmenta on face 3, and iouestod (iusto) on face 4.4 The date may be said to be fixed by the variation of the sign for m between and (with for r) and other alphabetic indications which suggest the 5th century B.C. It has been suggested also that the reason for the destruction of the stele and the repavement may have been either (1) the pollution of the comitium by the Gallic invasion of 390 B.C., all traces of which, on their departure, could be best removed by a repaving; or (2) perhaps more probably, the Augustan restorations (Studniczka, Jahresheft d. Österr. Institut, 1903, vi. 129 ff.).

42. Of the earlier long inscriptions the most important would be the Columna Rostrata, or column of Gaius Duilius (q.v.), erected to commemorate his victory over the Carthaginians in 260 B.C., but for the extent to which it has suffered from the hands of restorers. The shape of the letters plainly shows that the inscription, as we have it, was cut in the time of the empire. Hence Ritschl and Mommsen pointed out that the language was modified at the same time, and that, although many archaisms have been retained, some were falsely introduced, and others replaced by more modern forms. The most noteworthy features in it are—C always written for G (Ceset = gessit), single for double consonants (clases-classes), d retained in the ablative (e.g., in altod marid), o for u in inflexions (primos, exfociont = exfugiunt), e for i (navebos = navibus, exemet = exemit); of these the first is probably an affected archaism, G having been introduced some time before the assumed date of the inscription. On the other hand, we have praeda where we should have expected praida; no final consonants are dropped; and the forms -es, -eis and -is for the accusative plural are interchanged capriciously. The doubts hence arising preclude the possibility of using it with confidence as evidence for the state of the language in the 3rd century B.C.

43. Of unquestionable genuineness and the greatest value are the Scipionum Elogia, inscribed on stone coffins, found in the monument of the Scipios outside the Capene gate (C.I.L.1 i. 32). The earliest of the family whose epitaph has been preserved is L. Cornelius Scipio Barbatus (consul 298 B.C.), the latest C. Cornelius Scipio Hispanus (praetor in 139 B.C.); but there are good reasons for believing with Ritschl that the epitaph of the first was not contemporary, but was somewhat later than that of his son (consul 259 B.C.). This last may therefore be taken as the earliest specimen of any length of Latin and it was written at Rome; it runs as follows:—

honcoino . ploirume . cosentiont . r[omai] duonoro . optumo . fuise . uiro [virorum] luciom . scipione . filios . barbati co]nsol . censor . aidilis . hic . fuet a [pud vos] he]c . cepit . corsica . aleriaque . urbe[m] de]det . tempestatebus . aide . mereto[d votam].

The archaisms in this inscription are—(1) the retention of o for u in the inflexion of both nouns and verbs; (2) the diphthongs oi (= later u) and ai (= later ae); (3) -et for -it, hec for hic, and -ebus for -ibus; (4) duon- for bon; and (5) the dropping of a final m in every case except in Luciom, a variation which is a marked characteristic of the language of this period.

44. The oldest specimen of the Latin language preserved to us in any literary source is to be found in two fragments of the Carmina Saliaria (Varro, De ling. Lat. vii. 26, 27), and one in Terentianus Scaurus, but they are unfortunately so corrupt as to give us little real information (see B. Maurenbrecher, Carminum Saliarium reliquiae, Leipzig, 1894; G. Hempl, American Philol. Assoc. Transactions, xxxi., 1900, 184). Rather better evidence is supplied in the Carmen Fratrum Arvalium, which was found in 1778 engraved on one of the numerous tablets recording the transactions of the college of the Arval brothers, dug up on the site of their grove by the Tiber, 5 m. from the city of Rome; but this also has been so corrupted in its oral tradition that even its general meaning is by no means clear (C.I.L.1 i. 28; Jordan, Krit. Beiträge, pp. 203-211).

“The living Latin for all the higher forms of composition, both prose and verse, was a far nobler language than the living Greek. During the long period of Grecian pre-eminence and literary glory, from Homer to Demosthenes, all the manifold forms of poetry and prose which were invented one after the other were brought to such exquisite perfection that their beauty of form and grace of language were never afterwards rivalled by Latin or any other people. But hardly had Demosthenes and Aristotle ceased to live when that Attic which had been gradually formed into such a noble instrument of thought in the hands of Aristophanes, Euripides, Plato and the orators, and had superseded for general use all the other dialects, became at the same time the language of the civilized world and was stricken with a mortal decay.... Epicurus, who was born in the same year as Menander, writes a harsh jargon that does not deserve to be called a style; and others of whose writings anything is left entire or in fragments, historians and philosophers alike, Polybius, Chrysippus, Philodemus, are little if any better. When Cicero deigns to translate any of their sentences, see what grace and life he instils into their clumsily expressed thoughts, how satisfying to the ear and taste are the periods of Livy when he is putting into Latin the heavy and uncouth clauses of Polybius! This may explain what Cicero means when at one time he gives to Greek the preference over Latin, at another to Latin over Greek; in reading Sophocles or Plato he could acknowledge their unrivalled excellence; in translating Panaetius or Philodemus he would feel his own immeasurable superiority.”

Another example of what we may call the Latin of business life is supplied by Vitruvius. Besides the obscurity of many of his technical expressions, there is a roughness and looseness in his language, far removed from a literary style; he shares the incorrect use of the pluperfect, and uses plebeian forms like calefaciuntur, faciliter, expertiones and such careless phrases as rogavit Archimedem uti in se sumeret sibi de eo cogitationem. At a somewhat later stage we have, not merely plebeian, but also provincial Latin represented in the Satyricon of Petronius. The narrative and the poems which are introduced into it are written in a style distinguished only by the ordinary peculiarities of silver Latinity; but in the numerous conversations the distinctions of language appropriate to the various speakers are accurately preserved; and we have in the talk of the slaves and provincials a perfect storehouse of words and constructions of the greatest linguistic value. Among the unclassical forms and constructions may be noticed masculines like fatus, vinus, balneus, fericulus and lactem (for lac), striga for strix, gaudimonium and tristimonium, sanguen, manducare, nutricare, molestare, nesapius (sapius = Fr. sage), rostrum (= os), ipsimus (= master), scordalias, baro, and numerous diminutives like camella, audaculus, potiuncula, savunculum, offla, peduclus, corcillum, with constructions such as maledicere and persuadere with the accusative, and adiutare with the dative, and the deponent forms pudeatur and ridetur. Of especial interest for the Romance languages are astrum (désastre), berbex (brébis), botellus (boyau), improperare, muttus, naufragare.

Suetonius (Aug. c. 87) gives an interesting selection of plebeian words employed in conversation by Augustus, who for the rest was something of a purist in his written utterances: ponit assidue et pro stulto baceolum, et pro pullo pulleiaceum, et pro cerrito vacerrosum, et vapide se habere pro male, et betizare pro languere, quod vulgo lachanizare dicitur.

The inscriptions, especially those of Pompeii, supply abundant evidence of the corruptions both of forms and of pronunciation common among the vulgar. It is not easy always to determine whether a mutilated form is evidence of a letter omitted in pronunciation, or only in writing; but it is clear that the ordinary man habitually dropped final m, s, and t, omitted n before s, and pronounced ĭ like ĕ. There are already signs of the decay of ae to e, which later on became almost universal. The additions to our vocabulary are slight and unimportant (cf. Corpus Inscr. Lat. iv., with Zangemeister’s Indices).

69. In the vocabulary of Tacitus there are to be noted:—

1. Words borrowed (consciously or unconsciously) from the classical poets, especially Virgil, occurring for the most part also in contemporary prose. Of these Dräger gives a list of ninety-five (Syntax und Stil des Tacitus, p. 96).

2. Words occurring only, or for the first time, in Tacitus. These are for the most part new formations or compounds from stems already in use, especially verbal substantives in -tor and -sor, -tus and -sus, -tura and -mentum, with new frequentatives.

3. Words used with a meaning (a) not found in earlier prose, but sometimes borrowed from the poets, e.g. componere, “to bury”; scriptura, “a writing”; ferratus “armed with a sword”; (b) peculiar to later writers, e.g. numerosus, “numerous”; famosus, “famous”; decollare, “to behead”; imputare, “to take credit for,” &c.; (c) restricted to Tacitus himself, e.g. dispergere = divolgare.

Generally speaking, Tacitus likes to use a simple verb instead of a compound one, after the fashion of the poets, employs a pluperfect for a perfect, and (like Livy and sometimes Caesar) aims at vividness and variety by retaining the present and perfect subjunctive in indirect speech even after historical tenses. Collective words are followed by a plural far more commonly than in Cicero. The ellipse of a verb is more frequent. The use of the cases approximates to that of the poets, and is even more free. The accusative of limitation is common in Tacitus, though never found in Quintilian. Compound verbs are frequently followed by the accusative where the dative might have been expected; and the Virgilian construction of an accusative with middle and passive verbs is not unusual. The dative of purpose and the dative with a substantive in place of a genitive are more common with Tacitus than with any writer. The ablative of separation is used without a preposition, even with names of countries and with common nouns; the ablative of place is employed similarly without a preposition; the ablative of time has sometimes the force of duration; the instrumental ablative is employed even of persons. A large extension is given to the use of the quantitative genitive after neuter adjectives and pronouns, and even adverbs, and to the genitive with active participles; and the genitive of relation after adjectives is (probably by a Graecism) very freely employed. In regard to prepositions, there are special uses of citra, erga, iuxta and tenus to be noted, and a frequent tendency to interchange the use of a preposition with that of a simple case in corresponding clauses. In subordinate sentences quod is used for “the fact that,” and sometimes approaches the later use of “that”; the infinitive follows many verbs and adjectives that do not admit of this construction in classical prose; the accusative and infinitive are used after negative expressions of doubt, and even in modal and hypothetical clauses.

Like Livy, the writers of this time freely employ the subjunctive of repeated action with a relative, and extend its use to relative conjunctions, which he does not. In clauses of comparison and proportion there is frequently an ellipse of a verb (with nihil aliud quam, ut, tanquam); tanquam, quasi and velut are used to imply not comparison but alleged reason; quin and quominus are interchanged at pleasure. Quamquam and quamvis are commonly followed by the subjunctive, even when denoting facts. The free use of the genitive and dative of the gerundive to denote purpose is common in Tacitus, the former being almost limited to him. Livy’s practice in the use of participles is extended even beyond the limits to which he restricts it. It has been calculated that where Caesar uses five participial clauses, Livy has sixteen, Tacitus twenty-four.

In his compressed brevity Tacitus may be said to be individual; but in the poetical colouring of his diction, in the rhetorical cast of his sentences, and in his love for picturesqueness and variety he is a true representative of his time.

For the subsequent history of the language see Romance Languages.