29.1 Introduction

If there is one thing that emerges from this book it is that dividing the Latin language into neat chronological periods will not work without severe reservations. Usages that may seem to be ‘early’ often turn out not to be confined to a particular period, or alternatively their attestations may be genre-related, that is characteristic of a genre that happens to survive mainly from an early period. As for ancient grammarians and commentators on the language, no single concept of what ‘early Latin’ is may be extracted from their works. Early Latin, or the Latin of ueteres, was a different thing for different commentators. I would myself tend to use ‘early Latin’ (arbitrarily) of the Latin of the period before about 100 bc, provided that one excludes from the category ‘early’ usages which, though they were current early, also remained current beyond that time. One is looking for meaningful distinctions that can be set up between early and later linguistic practices, and never mind the precise chronological divisions.

I start with some remarks, based on a number of the chapters in this book, on the variability of these ancient attitudes.

29.2 Grammarians and Other Commentators

For Varro Garcea shows that ‘early Latin’ was a thing of the distant past that could be got at from picking up isolated facts about lost lexemes, obsolete meanings and earlier forms of words (Chapter 26). Varro attempted a systematic account of the linguistic changes that had taken place in the two hundred years before his time, setting out the causes of obscurity deriving from uetustas and the passing of time. These included the loss of lexemes, semantic change and the complexities of phonetic change, which required systematic scholarly interpretation. Significant traces of earlier Latin in Varro’s view are to be found particularly in the provinces. The linguarustica or Sabina is marked by a conservative character. For example, the rustic form ueha for uia confirms the word’s origin from uehere.

Holford-Strevens (Chapter 25) points out that for Gellius by contrast the ueteres lived before the time of Augustus, and included Cicero. A feature of their Latin was that it was ‘good’. Here is a definition, or a division, that brings out vividly the arbitrary nature of periodisation. There was also a group of uetustiores, from whom Cicero is explicitly excluded.

Different again was Valerius Probus, who lived in the Flavian period (discussed in Chapter 26 by Garcea). When his military career failed he devoted himself to reading authors still in school programmes in his province but considered out of date by the ‘fashionable literati of Rome’. These authors included Virgil, Sallust, Valerius Antias, Ennius and Plautus. Probus compares early and recent usages, but is non-prescriptive. Earlier Latin is not demarcated by a date or cut-off point, but embraces both the 2nd century bc and what we would call the classical period. Again we see an indifference to the sort of chronology that would typically be used by scholars today wishing to define ‘early Latin’. There is an implication that ‘fashionable literati’ were capable of deciding that fairly recent writers were now ‘out of date’, and what is ‘out of date’ (we might say ‘early’) becomes a shifting concept.

Different grammarians and antiquarians had their own motives in selecting words from the past to note or passages to quote, and their choices may be related not to a clear concept of what constituted ‘early Latin’ but to their particular purposes. Nonius Marcellus came from the town of Thubursicum in North Africa, a town with learned pretensions and an interest in the past. Welsh (in Chapter 27) shows that Nonius had interests somewhat narrower than those usually assigned to him. Welsh reviews some peculiar gaps in the collection of texts that Nonius consulted, and argues that these arose because Nonius consciously rejected certain types of texts. He consulted a large but eccentric collection of republican Latin literature. Very little oratory is represented. It is suggested, not that he was not interested in oratory, but that he was on the contrary particularly interested in public speech, and ‘wrote with an eye to the needs of contemporary public speakers’ (p. 553). When excerpting, Nonius followed the practice of Festus, in that he selected words from Pacuvius or Lucilius or Afranius that were fairly close to the patterns of Latin imperial speech, and intelligible without further explanation. These words are ‘on the bleeding edge of old-fashioned’, and not far removed from current Latin, for example, nouns with an unusual gender. He was selecting to help speakers who wanted to gain distinction in North Africa, and was perhaps seeking words, not part of the orator’s repertoire, that would be striking in contemporary public speech. Nor was he necessarily always reliable. For example, he inflicts non-Plautine meanings on Plautine words, in his desire to recommend an authoritative novelty without bothering about the precise meaning of the word in its original context.

In other words Nonius was more interested in his own period than in the early period for its own sake, and it would be a mistake to make generalisations about the character of Latin of an early period by taking everything in Nonius at face value.

Garcea notes that a conflict between prescriptive and anti-dogmatic linguistic attitudes emerged. There was an awareness that linguistic evolution was relentless, and grammarians such as Palaemon promoted a prescriptive method in school education, which offended the non-dogmatic, such as Quintilian.

At the same time an archaising movement was developing, and Quintilian had to face a problem: the application of analogy produced monotonous standardisation, whereas appeal to auctores and the past could offend the linguistic notions of the public. Quintilian rejects analogy and produces a new set of criteria (1.6.1 sermo constat ratione uetustate auctoritate consuetudine). Consuetudo ‘customary usage’, should be based on consensus eruditorum. Archaisms were prestigious, but must be used sparingly, to avoid obscurity: see Inst. 1.6.39–40: ‘moderation is essential’. Words taken from antiquity (a uetustate) ‘must not be taken from remote and now forgotten ages’. It is not a matter of reviving old words from the distant past. ‘Early Latin’, or uetustas, is left vague, but it is not on the whole usage from a long time ago. On the other hand Quintilian does say (Inst. 1.6.1) that ‘antiquity is, if I might almost say, commended by religious awe’ (religio). Later (Inst. 1.6.40) he says that religion forbids us to change certain words. There is perhaps implied here an acceptance of religious archaisms. Quintilian does not express a clearcut concept of ‘early Latin’ and its dates, and is interested not in antiquity/ancient usage as a subject in itself, but in maintaining comprehensibility in oratory: see Inst. 1.6.41 ‘But how faulty oratory (whose chief virtue is clarity) would be if it needed an interpreter.’

Particularly interesting are the fragments of Caper, who is placed between Probus and Julius Romanus, the latter of the second half of the 3rd century ad. Priscian produces two lists to do with gender variation, the first comprising variation between masculine and feminine, the second between neuter on the one hand and masculine or feminine on the other. The lists derive partly from Caper. The great majority of the examples come from the period before the 1st century bc, but there are some examples from for example, Varro, Catullus, Lucretius, Virgil, Horace, Caesar, Livy, Ovid and even Lucan. This is the work of a scholar surveying a long period of the language’s earlier history, but the predominance of the early period (pre-1st century) cannot be put down solely to greater variability of gender early on, because gender variation continued indefinitely (see Adams Reference Adams2013, Chapter xix). Caper seems to have had an interest in the earlier period for its own sake. A second fragment of Caper, also from Priscian, this time to do with active uses for passive or deponent for active, shows a very similar selection, though could it rather reflect in this case the state of the language at different times?

The variability of later attitudes to what constituted early usage is illustrated by Garcea from two passages of Priscian. In the case of scidi/scicidi Priscian distinguishes between modern writers and early ones by putting in the latter category Afranius, Accius, Naevius and Ennius, whereas in the case of pellicui/pellexi the usage of ueteres is illustrated from more recent authors, Laevius, Varro Atacinus, Calpurnius Piso.

Damon’s chapter (24) is about the use of Cato by Pliny the Elder and Pliny’s rewriting of what he found in the source. She shows for example that Pliny replaces future imperatives in passages that he took over from Cato. However, at one point (Nat. 18.315–36) Pliny uses material from other sources, adopting Cato’s style, notably the constant use of the future imperative. He knew how to write like Cato but chose not to most of the time. Here is another reason why simplistic periodisation of the language is not convincing: we cannot say that the future imperative had fallen out of use between the time of Cato and of Pliny. With the passage of 200 years the language may retain a certain resource or usage, but individual attitudes to it may vary, causing apparent ‘language changes’ that are really only matters of fashion. As Damon puts it, Pliny was writing in a different genre.

One cannot therefore attempt a definition of ‘early Latin’ by examining the attitudes of ancient commentators to their predecessors. These attitudes are diverse, and reflect the fact that different individuals had different reasons for commenting on usages they had encountered in earlier texts.

29.3 The Problematic Character of Some Evidence

It may be the case that spoken languages change gradually and that with modern methods of data collection it is possible to plot changes chronologically, but written evidence is another matter. Much of the evidence for Latin belongs to certain literary genres, and literary genres do not simply abandon the past. There are bound to be usages shared by Ennius on the one hand and Virgil on the other in epic poetry, and their persistence over time in that genre inevitably obscures changes in speech and makes the setting up of chronological stages based on extant literature highly dubious. The same reservation applies to late Latin, because so many late texts are influenced by biblical Latin and have biblical quotations and allusions embedded constantly.

A major problem arises from the changing nature of the evidence surviving on the one hand in the period down to, say, 150 bc, and that from later. Pezzini (Chapter 11) addresses this and other problems in considerable detail. He starts with a corpus of c. 1220 words that are only attested in ‘early Latin’ (his end point for ‘early Latin’ and for the start of ‘classical Latin’ is the start of Cicero’s career; that in itself is a dividing line that raises questions), but then shows that many of these, for various reasons, have to be ruled out as unequivocally ‘early Latin’ (i.e. as used in the early period only but then dropped from the language). I start with his third category. In the early period a lot of the texts that survive are comedies, but little in the way of comedy survives from the classical period. It follows that terms attested only early (in comic sources) may still have been in use later in ordinary speech of the type which comedy, up to a point, undoubtedly drew on, but are submerged later on because of the chance survival of some types of texts but not others. Pezzini includes among usages in this class interjections and low-register conversation markers.

One of these is edepol, which is the object of a detailed study by Vine (Chapter 10). This Vine analyses as comprising three early elements: there can be no doubting its antiquity. Its distribution is at first sight in favour of its ‘early Latin’ credentials, in that it occurs almost 390 times in Plautus and Terence but thereafter only about 20 times, mainly in republican drama. But as Vine puts it, the term was ‘at home in comic drama, with virtually no afterlife once that genre disappears from the literary record’ (p. 210). The problem of course is that it might not have disappeared in speech, and that the decline in its attestations may be due entirely to the chance character of the surviving literary record.

Another inadequacy of the evidence is that many words that occur only or mainly in the early period, however one defines that, are, as Clackson (p. 396) says, ‘absolute hapaxes’. A word that occurs just once in a period may be of very limited linguistic significance. For example, a Greek word introduced by an early writer in a single passage may not reflect the state of the language at the time but have been an ad hoc borrowing by the writer motivated by any number of factors, such as word play, learned reference or a momentary language shift, that is a code-switch. According to Clackson, of the Greek loanwords or derivatives of loanwords found in Plautus, only four are attested more than forty times, and all four words are outside the 800 most frequent words in Plautus.

Pezzini too (Chapter 11) deals with hapax legomena. A hapax need not be a Greek loanword, and there are no doubt terms of many categories that occur just once, for example motivated by an unusual subject matter in a particular context. But one factor that certainly undermines the significance of mere lists of terms confined to ‘early Latin’ is the habit of the comic poets of coining comic neologisms, which never caught on in general usage. Pezzini is careful to eliminate such material in his assessment of the ‘early Latin’ lexicon.

Pezzini also eliminates as possible ‘early Latin’ lexical items various special categories that need not have been genuinely restricted to the early period, or need not, despite appearances, have been genuinely current in the early period. His corpus of c.1220 words potentially construable as ‘early Latin’ is reduced to a corpus of just 204 early terms after necessary eliminations.

A lack of genuine currency of a different type, displayed by a group of usages in the period down to about the mid 2nd century, is stressed particularly by de Melo (Chapter 6), namely ‘line-end archaism’. For example, over 90% of old forms of esse (such as siem) in both Plautus and Terence occur at the end of lines. A line-end archaism is an archaism used predominantly for metrical reasons. In a prose text, such archaic forms would be considerably rarer. What is also shown by such a distribution is that Plautus and Terence obviously had a concept of the archaic and of linguistic change. Maltby (Chapter 12) confirms the status of siem as an artificial type of archaism. He produces figures showing the comparative frequency of for example, sim versus siem in early comedy and tragedy, and the reduced forms are much preferred in both genres. If siem were simply an archaic styleme it would be expected to be proportionately more frequent in the higher genre tragedy. It was clearly out of use and used for metrical reasons, not as an old-fashioned word form.

We do however in this way catch a glimpse of a usage that by definition pre-dated the time of Plautus. Another such form dealt with by de Melo (pp. 103–5) is the passive infinitive of the type amarier (in contrast to amari). The rarity of such forms even as line-end archaisms in Plautus and Terence shows that they were well on the way out, and predictably in classical prose are almost non-existent. Line-end archaisms are suggestive of chronological stages in the early Republic itself. De Melo also finds evidence of a different motivation for some early forms. The ai genitive, as in magnai rei publicai gratia, is used very deliberately, as a marker of higher register.

So far, as possible patterns indicative of ‘early Latin’ features, I have alluded mainly to what appear to be dramatic changes in the incidence of certain usages between the early period and later periods, and it has been suggested that in many cases, but not all, declining frequency is illusory and has to be explained as due to the vagaries of the survival of texts. There is, however, another possible pattern that may throw some light on ‘early Latin’. I refer to usages which, though attested throughout Latin, appear to have a higher incidence in the early period. Baños studies ‘support verb’ constructions over a long period, with statistics (Chapter 7). Support verbs in their entirety cannot from his table be attributed especially to ‘early Latin’. They have their highest incidence in Sallust, and in other historiographical texts. On the other hand Plautus’ and Terence’s marked preference for esse as a support verb is a salient feature of ‘early Latin’. Here we see a usage that continues throughout Latin but declines after the early period. It is not unique to ‘early Latin’, but its high comparative frequency is. The pattern showing a verbal noun in -tio combined with esse is much more common in ‘early Latin’. Baños employs the term ‘generic’ support verb constructions in reference to those with esse, facere, dare and habere. These make up the highest proportion of support verbs in Plautus, and also a high proportion of such constructions in Terence. The statistical significance of such constructions declines later but they continue to be in use. Once again, however, any attempt to find a straightforward chronological change from the early period to the later is undermined by a striking exception to the general point just made: the Peregrinatio Aetheriae, of the 4th century ad, has the highest proportion of all. Again we see the possibility that a usage had not been dropped from use but had become submerged.

The inadequacy of simply compiling statistics for a phenomenon in several different periods is nowhere made clearer than it is by several figures from Clackson’s chapter (19). He counts 4295 Greek words listed in the OLD, but 1363 of these occur only in Pliny the Elder. Of the 2932 words left from the OLD, 76% are first attested after the early Latin period. Clackson adds that ‘large areas of technical and scientific language … are represented scarcely or not at all in the early Latin period’ (p. 401). The figures are determined not by changes in the language, but by the different character of the corpora surviving in the various periods. Clackson also shows that, ‘periods’ aside, Latin has a low level of borrowing from Greek compared with the amount of borrowing in modern languages, as shown in the World Loanword database. The language was not ‘flooded’ by loans from Greek in comparative terms at any period. For example, despite the notion that Greek loanwords were widespread in spoken Latin, in Plautus, supposedly a major source for everyday speech, there are only, as was pointed out earlier, four Greek loanwords or derivatives of loanwords that are attested in the plays more than forty times (pp. 396–7). It is also the case that many Greek loans in Latin are hapax legomena, and not therefore necessarily in proper use in Latin.

29.4 Aspects of ‘Early Latin’

We have stressed the problems caused by any attempt to divide the language into an ‘early’ period and a later, but that does not mean that early features of one sort or another cannot be identified. Changes, with different chronologies, can up to a point be identified. The word ‘change’ itself in reference to written language use may however have different meanings. ‘Language change’, for example, need not apply to the whole language, but refer to shifting fashions.

Some evidence about lexical and some other usages in the early period (down to about 150 bc) will tell us about generic differences in that period but not about general features of the period in comparison with other (later) periods. Generic differences do however vary in their significance. If pauor is mainly a word of early tragedy rather than of comedy (see p. 256 in Chapter 12 by Maltby), that does not tell us much. The word is common later in all sorts of genres, particularly denoting the fear of large groups. An important determinant of its use was semantic. On the other hand a generic distinction in lexical or other linguistic choice in the early period may itself constitute a feature of the early period if at later periods that generic distinction did not persist. If for example asyndeton bimembre in some legal and official documents before the Ciceronian period is more common than explicit coordination, whereas in the 1st century ad in some similar documents it is far outnumbered by coordination, then there has at least been a change of fashion in one type of text, even if the language in all its varieties had not dropped asyndeton bimembre (see pp. 294–6 in Chapter 14 by Adams and Nikitina). It is essential however to make a distinction between changes of linguistic fashion in particular types of texts, and such changes in a language in general. As for asyndeton bimembre, though it is possible to find laws in the post-Ciceronian period in which it is avoided, in later legal texts such as the Digest it remains common. Asyndeton looks like a form of legal shorthand that appealed to some but perhaps not to others, and its distribution is constantly variable.

There were possibly similar changes of fashion in religious language too. In the prayers examined by Adams and Nikitina asyndeton bimembre is rare, but there is an interesting example in an augural context at Pl. As. 259–60 that is very similar to an example in the same sort of augural context in Umbrian (Tab. Ig. VIa.1). The formula must have been very old, and the possibility must be allowed that the rarity of asyndeton in prayers reflects changing fashion.

Generic differences in the early period may on the other hand sometimes be relevant to the history of the language in a wider sense. The old masculine plural genitive ending -um tends to be used in certain collocations, particularly in comedy (in Plautus, but not Terence, who hardly has it). There is a particular interest to Maltby’s collection of such forms in the tragic fragments of Pacuvius. They are numerous, and not restricted to formulaic collocations. Maltby concludes that in tragedy they were used to elevate the stylistic level whereas in Plautus they ‘seem not to be used for any particular stylistic effect’ (p. 265). The form must have fallen out of general use but still have been available as a remembered morpheme of elevated tone. Later its use changes again. Tacitus, for example, does not use it as a styleme, but largely restricts it to certain words in which it had retained a formulaic status. Cicero does exactly the same, and interestingly he describes his own practice at Orat. 156.

New Latin inscriptions from the early period are rare, and that makes eleven bronze rostra from warships found since 2004 off western Sicily, near the Egadi islands, potentially interesting (see Chapter 4 by de Melo). The names of officials are given, with abbreviated first name, family name and abbreviated filiation, followed by a form of probare in the perfect. The probatio is an indication by an official that the rostra had been produced in the proper way. One person is identifiable, Marcus Publicius, aedile in 241 or 238, consul in 232 bc. He is quaestor here, and the date must be in the 50s or 40s of the 3rd century. Since one inscription is in Punic, the ships in question were sunk in the First Punic War, and that must have happened in the major battle off the Egadi islands on 10 March 241 bc. It is thought that the most likely date of the inscriptions is 250 or 249, or perhaps 254 bc.

The inscriptions are marked by the form Papeirio alongside Paperio (originally Papeisios). The presence of the original diphthong and also its replacement is of potential importance as an indicator of the state of the language. De Melo (p. 69) refers to Wachter as quoting an e spelling from ei from a Roman inscription of exactly this date, 241 bc. Similarly (p. 72) probauet derives from -eit, again with monophthongisation (to e) but not yet closing to long i (before shortening in final syllable). This along with Paperio (in the absence of Papirio) suggests that ei > e > i had reached the intermediate stage, with the retention of ei a spelling possibility. The later tradition of recommending the writing of ei is a spelling rule unrelated to pronunciation. It is true in general to say that the spelling of the language is not a reliable indication of the state of the language itself. The word ‘archaic’ as applied to mainly old forms is ambiguous, as it may refer to the state of the language, or it may refer to an artificial cultivation of the old-fashioned. The evidence from the rostra, with ei coexisting with e but no sign of i, does however suggest the state of the language in this respect. Later those who write ei as an old-fashioned form were well capable of lapsing into i. Ei and e spellings go on much beyond the early period, but as spelling artificialities. What makes the rostra interesting is the absence (so far) of i.

One does not expect in language history sudden radical changes. Changes are gradual, with long coexistence of the old with the new. Statistics may reveal such gradual developments. Barrios-Lech (Chapter 9) examines the mood in indirect questions in Plautus, Terence, the Annales and tragic fragments of Ennius, and in Cato’s de Agri Cultura, and finds 1636 indirect questions in total. He concludes (pp. 204–5) that there is ‘no firm evidence for the claim that the subjunctive gained ground on the indicative in interrogative clauses of indirect question from the early to the classical period’. Both moods continued into the classical period, but there was a decline between Plautus and Terence, that is, even in a quite early period, in the incidence of the indicative. The uiden ut (incedit) type, for example, disappears from all but high literature by the classical period and had probably fallen out of use. Barrios-Lech notes that Terence’s most ‘Plautine’ play is the Eunuch. Viden ut is excluded by Terence from all plays but the Eunuch.

The study of metre in Plautus and Terence on the one hand and the later period on the other throws some light on changes in the language, as de Melo and Pezzini (Chapter 5) show. For example Plautus and Terence preserve some final geminate consonants, as in miless (from -t-s) at Pl. Aul. 528, but in classical Latin, though hoc still scans as hocc, degemination has taken place in polysyllables.

Pezzini’s identification (Chapter 11) of categories of ‘early Latin words’ that are only superficially so has the positive consequence that after the painstaking elimination of the many false early words some terms do remain that are not susceptible to such elimination. Many terms in Terence, for example, can be eliminated, but after the process of elimination there do remain twenty-seven words that might be genuinely ‘early Latin’. Actutum is a case in point, which goes from very common in Plautus, to rare in classical Latin, where it turns up only in stylised contexts. Pezzini lists a variety of such words.

But must every word or morphological form that is attested have once been in standard use, even if its distribution in the recorded period is exclusively typical of high, artificial style? Take the adverb topper (see Goldberg, p. 278 with n. 13). This was found in ‘ancient writings’ according to Nonius, and it turns up in a few high-style early poetic works and occasionally in similar contexts in the classical period. If in the pre-historic period it had been widely current but by the 3rd/2nd century bc was very much a stylised archaism, one could say that a prehistoric change had occurred, such that it had fallen out of ordinary use and come to be used only as a revivalism in high-style poetry. But we do not know whether it had once been widely current. Its etymology is unknown, and, since it means ‘quickly, at once’, it had synonyms of more familiar type. It may always have belonged to higher registers. One cannot assume that the prehistoric language was without these. If it was always of artificial character then the language had not changed at all.

One must therefore look carefully at apparent revivalisms in early poetry when assessing whether they do in fact reflect an earlier stage of the language. I would make a distinction between for example, topper and the -as genitive singular used by Andronicus and Naevius (see Goldberg p. 280 n. 18). We know from comparative evidence that this was the earliest form of the genitive in the first declension, and that it was replaced by -ai and subsequently -ae. It must have been in use once, and a change had clearly taken place in the language by the 3rd or 2nd century bc. As Goldberg puts it, when Naevius ‘describes the giants Runcus and Purpureus as filii Terras, his very language … thrusts the reader back into a long vanished time’ (p. 278).

Reduplicated perfects of the type tetuli = tuli are another usage that must genuinely have been current in the pre-Plautine period, to judge from some features of distribution commented on by Maltby (p. 254). In Plautus tetuli outnumbers tuli, whereas in Terence it is tuli that is regular. Donatus comments on three reduplicated perfects in Terence, which are all used by old men. There is a hint here that the form was falling out of use between the time of Plautus and Terence, except that it retained some currency in old-fashioned speech of the type associated with the elderly.

29.5 Aspects of Reception

Classical and later writers from time to time use Latin of an earlier period for particular reasons. Old phrases may be remembered but if a later user is not familiar with the contexts in which the phrases were originally used modifications may occur. Not all modifications need be modernisations either. A case in point is the change to the use of the old religious pair uolens propitius. In its earliest manifestation it is not an asyndeton bimembre, but later, for example in Livy, it is given different verbal complements (notably adsis for the earlier sis), and this has the effect of converting it into an asyndeton, a modification that would probably have been regarded as suited to an early period because of a sense that asyndeta bimembria were often old and formulaic (Adams and Nikitina, p. 295).

Another old phrase with two forms is populus Romanus Quirites versus populus Romanus Quiritium, commented on by Briscoe (p. 468). What looks like an early asyndeton is for example in a fragment of Fabius Pictor ap. Gel. 1.12.14 sacerdotem Vestalem facere pro populo Romano Quiritibus. An alternative formulation has Quirites in the genitive plural dependent on populus Romanus, as for example at Livy 41.16.1 and 22.10.3 donum duit populus Romanus Quiritium. According to Varro (L. 6.68) the Quirites were the Curenses or ‘men of Cures’, who came with king Tatius to receive a share of the state (see de Melo Reference de Melo2019: 2.395). On this view the Quirites would originally have been the non-Roman element in Rome, and the original phrase may have been populus Romanus Quirites, an asyndeton bimembre denoting the Roman and non-Roman original inhabitants of the city (for the type of asyndeton cf. patres conscripti, which originally referred to two groups, patricians and those drafted into the senate to make up numbers: see Adams Reference Adams2021: 252–4). If so the form with genitive Quiritium, used for example in an archaising context in Livy, would have been a modernisation, derived from false analysis of the original asyndetic phrase in the genitive throughout (populi Romani Quiritium in the genitive > populus Romanus Quiritium in the nominative).

It is one thing to get an old word wrong in some way, and another to reinvigorate it by using it (deliberately) in a new way. Taylor (Chapter 21) illustrates some usages of this type from Lucretius. At 1.619 for example Lucretius takes up the form escit, which had been used in the Twelve Tables as the existential correspondent of the present tense copula est, and not only imposes on it a future meaning (it is picked up by erit in the next line), but also uses it as a copula verb. It happens to suit the metre to use it at 619 rather than erit, but then Lucretius was not constrained at 619 to produce a line that had the verb ‘to be’ at the end, and it is clear that he has used the form inventively while introducing an archaic note. It will be remembered that Nonius, as Welsh shows, was happy to ‘inflict non-Plautine meanings on Plautine words’ (p. 584): he did not bother about the precise meaning of some words in their original context. Taylor gives other examples of the same phenomenon from Lucretius. For instance, at 1.71 Lucretius uses cupiret with fourth conjugation morphology, a deviation from the norm that can be paralleled in type in ‘early Latin’ (Ennius, Plautus, Cato), but not in this particular verb.

I return finally to genre. It was remarked earlier (section 29.3) that literary genres do not abandon the past. Sometimes the old may continue in a particular form of composition even if it is out of line with current usage, and that is another reason why setting up chronological stages may be pointless. This is demonstrated in detail in Panayotakis’ chapter (20). Two of the writers of Atellane farce, Pomponius and Novius, were nearer in time to Cicero than to Plautus and Terence, but Pomponius in particular seems to have written in the style of Plautus and Terence (note, for example, adverbs with the suffix -iter, such as parciter for the usual parce). His language was that of a timeless comic genre, not of a period. As Panayotakis suggests, ‘early Latin’ transcends chronological boundaries.

29.6 Conclusions

Latin is attested over many centuries, and it was definitely not static. There was not however an entity ‘early Latin’ in use until a convenient date (e.g. 150, 100 bc), which then changed into ‘classical Latin’. I leave out of consideration here the period of the language sometimes referred to as ‘proto-Latin’, that is from the time of the kings, about which classical writers knew little or nothing. The genuinely early in our more restricted sense consists of morphological features (such as siem for sim, the -as genitive, the -um genitive plural, reduplicated perfects such as tetuli), or spellings indicative of pronunciations that had passed out of use by the late Republic (e.g. e for ei), phenomena which are either in inscriptions such as the Egadi rostra or are buried away in early literary texts in special contexts (e.g. at the end of lines or in formulae or in genres such as prayers given to the use of archaisms). What we do not see from such evidence is a living early language in which these phenomena were the norm. A writer who admits a line-end archaism is providing us with evidence of the former existence of such a form, but not of the way the form had been used in this imagined earlier stage of the language. Was it the norm in the earlier period, or was it stylistically artificial even then and in competition with an equivalent form? Miscellaneous pieces of information about ‘early’ usages provided by grammarians and other commentators suffer from the same limitations.

Various approaches may be adopted to recover early phenomena. First, statistics may be compiled for the incidence of a phenomenon not simply in the few genres of Latin before a certain date (say, 150 bc), but for a whole range of genres over a much longer period. Subtle distinctions may emerge between the features of the phenomenon in the early period compared with those of later periods. That is the approach adopted by Baños (Chapter 7) in his study of support verbs. Second, distributional features of a phenomenon in early texts down to about 150 bc may throw light on the existence of the phenomenon even earlier and its status by the time it is attested. Comparative evidence may also suggest that a form (e.g. the genitive plural -um) had once been a presence in very early Latin. Third, there are represented in early texts or at least republican texts several literary genres composed by different practitioners at different dates, and comparisons of certain phenomena in the earlier and the later works may show change in operation and thus highlight an ‘early’ feature. Barrios-Lech (Chapter 9) for example shows that one or two uses of the indicative in indirect questions fade in comedy from Plautus to Terence. Fourth, one may make a distinction between a feature that is early in a strict sense, and a feature that is particularly ‘fashionable’ early, or at least fashionable in one genre in the early period, more so than in later periods. For instance, end-of-list coordination is a marked feature of early prayers (Adams and Nikitina, pp. 298–301), but it is not common in comedy of the early period, or in a range of literary genres in classical Latin, though it continued to exist and eventually was to become the norm in Romance languages. It was a vogue usage in a particular genre in early republican Latin.

In conclusion, this volume has interpreted ‘early Latin’ liberally, as embracing the period from some very early inscriptions down to about the first quarter of the 1st century bc, from which time there are writings, literary and otherwise, that still have marked affinities with the earlier language. ‘Classical Latin’, that is Latin from later decades of the 1st century into the early Empire, cannot of course be divorced sharply from the language of the earlier period, but it offers serious problems of definition related to the notion of ‘standardisation’, which ideally should be addressed in a volume of its own. Classical Latin has thus found a place in this volume, but in a very restricted sense. In the classical period (and beyond) there were writers who tried to imitate or reproduce in places the Latin of an earlier age, and there were also interpreters of early Latin, who commented on some of its features. These later writers are relevant to the establishment of earlier texts, and they cannot be disregarded in any proper assessment of the early period. The section on ‘Reception’ deals with some of these later writers and their views, whether expressed directly or implied, on early Latin language. This section has been necessarily selective, and there has been no systematic attempt to cover ‘archaism’ in its variety of literary practices and functions from, say, Sallust to Apuleius (two important absences in this volume). This topic, too, should be addressed in a separate study, which, it has been our hope, the present volume may encourage.