The phenomena that come up in this anthology are far too diverse to be summarised, but there are some recurrent themes, which I will attempt to bring out here, not discursively but referring to discussions in the commentaries.
1 Periodisation
The texts in this work span more than 1,000 years, and one cannot but say something about ‘periodisation’, that is the question whether the language can be divided into clear chronological stages. This volume is however an anthology, and any selection of texts is bound to be a weak basis on which to build narratives of chronological change. Indeed the whole of extant Latin literature is an anthology, selected by scribes for copying in line with obsessions of the medieval period. That is why we have so much Cicero and so little Varro, so much Livy and Tacitus but almost nothing of their sources, such a huge amount of Christian writing from later antiquity but relatively little pagan, and so much high-style writing in both prose and verse and so little practical or personal writing on mundane subjects. There is no point in trying to define neat periods in the history of the language from such data. I will say something here about some different periods (loosely defined) in which Latin is extant, but not with the aim of identifying stages in an even progression. That is not to say that the language did not change, but rather that false conclusions can be drawn from imperfect evidence treated as offering the whole truth.
There are quite a few early texts (from about 200 to 160 BC) in our selection. ‘Archaic’ is sometimes used of the Latin of this period, but it can be an unsatisfactory term, implying for some readers that there was a ‘primitive’ stage in the history of the language. There is nothing primitive about early Latin. It is diverse in style and register, with subtle variations in individual writers. The fragments of Ennius’ Euhemerus at first sight look crude, with verbal repetitions, constant use of temporal adverbs and redundant use of is, but there are also signs of stylistic artificiality (see 1 Conclusions). Ennius did not have to write like this, as other forms of his output show. Terence (see 1 Appendix 3) and others, including annalists (see e.g. 1.8), sometimes used the same simple narrative style (for Plautus see Courtney Reference Courtney1999: 153–5), but Plautus when he wanted wrote elaborate narrative, abounding in artificial usages, as in a long passage of the Amphitruo (see 1 Appendix 4). The ‘crude’ style is not a sign of the (archaic) times, but a traditional method of narrating events simply. It was associated with folk tales (see 1 Appendix 3 and the first passage of Terence cited there), and may have had input from Greek. Nor did it die out after the early period, but was always available (see e.g. 8.3–4 and index, ‘popular/simple narrative style’).
There are also generic differences in early Latin texts, which tell against any notion that the language was naive. Notable phenomena that have come up are the use of atque before consonants (see 1.6 and 1 Appendix 1), which is all but non-existent in Cato’s De agricultura but constant in the fragments of his Origines, the use of the archaic form -ere of the third person plural perfect (1.11), and the use of present participles governing an accusative object (1.15). For ‘register variation’ in Plautus see 3 Conclusions.
Texts of the early period do of course have features that largely disappear from use later, such as the uninflected future infinitive form (see 6.14 on daturum); this and various such usages have been used to give a rough date to the Johns Hopkins defixio (see 6 Conclusions). Others in other texts include clam + accusative (1.16), gestio in its etymological sense (2.7–8), facinus in a neutral meaning (3.418), enim in initial position (3.429), lien ‘spleen’ (5.27), early subjunctive forms of edo ‘eat’ (5.19), and ualidus with an early meaning (4.13). Such phenomena in Plautus are listed at 3 Conclusions, ‘Old Latin features’.
But the Latin of texts of the early period was not a discrete entity. We have seen anticipations of later, including Romance, phenomena, as in Cato’s use of unde (4.5), of prepositions in various types of temporal expressions (5.6), of subtus (5.8) and pullus (4.10), and of sanum (facere) with the adjective showing a lack of concord with the primary (nominal) object of the verb (5.15). In Plautus analytic intensive superlatives are as common as the synthetic type (see 2.24), and although the intensifying element is variable and there is no sign yet of grammaticalisation, his taste for the combination intensifier + positive adjective provides the background to later developments. The verb fabulor, which he has often, is mainly submerged later but it continued into Romance languages (3.424). The reflexive dative of advantage, which is sometimes pleonastic, was to have a long history (2.5). Partitive expressions comprising de or ex standing as object of a verb are already found in Plautus and Cato and anticipate the Romance partitive article. The (apparent) accusative pronoun form ecillunc (in the Johns Hopkins tablet) is closer in its syntax to Romance compounds with ecce than are comparable forms in Plautus (see 6.43). Some (fading) usages lasted well beyond the early period, particularly but not exclusively in colloquial varieties of the language, such as the adverb istic (3.421), the reinforced pronoun istic, notably in the nominative and accusative singular forms istic and istunc (see 3.420, 8.1, 17.3), and the adverb illi (9.16). In the Conclusions to the section on Vitruvius (11) we have included a section entitled ‘Vitruvius and early Latin’, the point being, not that Vitruvius was an archaiser but rather that there is not the sharp division between the early and late Republic/Augustan period that there might appear to be if one uses only Cicero for defining ‘classical Latin’.
The classical period (which is taken to mean here the late Republic and early Empire, down to about the mid-first century AD) suffers from its persuasive adjective ‘classical’, and from at least one other term applied to it (‘Golden’ Age, as distinct from ‘Silver’). I would stress the limitations that have emerged to the standardisation that was supposedly effected during this period; a ‘standard language’ is readily given adjectives such as ‘classical’, since it is regarded as controlled. Here is some relevant evidence.
Cicero has not been included in this anthology except in the section on jokes, but he does come up indirectly. In a number of places it has been pointed out that usages found in his (educated) correspondents (or other contemporaries, notably Varro and Vitruvius) are used by Cicero himself either not at all or hardly ever. I would conclude from this that Cicero’s Latin cannot be taken as wholly representative of the late republican educated language. It is, inevitably, an idiolect. Because the works of Cicero survive in such abundance, they have traditionally been used to construct a classical Latin that obscures the diversity of the period. Late republican or Augustan deviations from what appear to be Ciceronian norms are traditionally classified as ‘vulgar’, but that is a tendentious classification when they might just as well in at least some cases be treated as evidence for diversity. See 9.9 (peream si), 9.10 (the adverb istoc), 9.11, 13.10 (discupio), 9.16 (the adverb illi), 11.5 (passive use of persuadeo), 11.5 (ab expressing agency with some abstracts), 11.16 (est + infinitive: here Cicero is at variance with a correspondent, and with Varro and Vitruvius), 11.17 (pluperfect of habeo for imperfect), 11.18 (impero + infinitive), 14.4 (mihi crede); see too 22.12–13, on the lack of standardisation in the forms of the perfect of compounds of lego. It is also possible that some ‘abnormalities’ in Cicero have been emended from his text (see 11.26 on the fifth declension genitive singular; another case is the indicative in indirect questions: see Adams Reference Adams2013: 752–3). It is unfortunate that so little is extant of Cicero’s contemporary Varro, who was famous for his learning. What does survive is very different in some ways from the language of Cicero (see Laughton Reference Laughton1960, and the index to the present work, s.v. ‘Varro, Latin of’).
Cicero’s Latin in the speeches changes over time. Usages found in the early speeches are sometimes dropped or reduced in frequency later (see e.g. 1.3 ibi tum, 3.421 istic, 3.425 homo with indefinites, 6.43–4 disperdo, 7.5 sis ‘please’, 7.8–9 compounds in per-, 9.8 nisi si, 14.7 pusillus, 16.2 ausculto). Such modifications suggest that there was at the period some debate about or at least consideration given to the acceptability of various usages, but any such debate implies stylistic awareness and not necessarily anything so dramatic as a deliberate movement that might be labelled ‘standardisation’.
Genre and register, and not merely personal taste, are of course major determinants of diversity in the classical period. So it is for example that usages peculiar to or at least mainly found in epistolography (including the letters of Cicero himself) have often come up. See e.g. 2.24 ualde, 3.417 opinor, 3.420 istaec, 3.421 istic, 3.428 timeo/timor, 6.29 adiuto, 7.5 heus (tu), 7.13 oricula, 9.10 isto, 9.13 hui, 12.3 bucca, 22.24 illim.
Another writer not dealt with explicitly but who is sometimes mentioned here, usually as exemplifying abnormalities, is Livy. The question arises why these should appear in his text.
In the Conclusions to 12 it is noted that Augustus in a fragment used ab with the name of a town, an exemplification of a practice of his commented on by Suetonius. This is a usage that Livy has constantly (Adams Reference Adams2013: 329–30), invariably writing ab Roma. Nor is it absent from other writers of the classical period (e.g. Sallust), though not frequent. The rule that prepositions are not used with names of towns has hardened into something immutable in the school tradition, but the reality is that practice was more varied, with Augustus and Livy consciously, it seems, taking a decision to use prepositions (Augustus for clarity, we are told). It was also noted that in Livy prepositions start to appear with domus, in contexts in which most Latin of the classical period would have a plain case (46.4).
Nisi si for nisi is shared by Livy with e.g. Cicero’s early works, Varro and Vitruvius (see 9.8). Fabulor occurs just once in Augustan literature (it is avoided by Cicero), in a speech in Livy (3.424). The adverb istic occurs twice in Livy, both times in speeches (3.421). The redundant use of homo with indefinites is found in speeches in Livy (3.425). Est + infinitive is in a speech in Livy (11.16). The ellipse of tam after a negative in the tam … quam construction is usually in direct speeches (22.15–16).
Livy famously was accused by Asinius Pollio of some sort of provincialism (Patauinitas: see Adams Reference Adams2007: 147–53 for details), and the presence of some of the above usages in his work does make one wonder whether they might be the sorts of things that Pollio had in mind (see 3 Conclusions). However, a number occur in speeches, and historians did admit informal usages in speeches. By contrast Livy’s examples of ab Roma are common in the narrative, and are suggestive of a lesser degree of standardisation than is allowed by the grammatical tradition. A more systematic search for such abnormalities both in speeches and narrative might turn up a firmer basis for Pollio’s remark, or, alternatively, further evidence for the limits of standardisation.
Thus, the common narrative of Latin periodisation that would have the language moving towards standardisation or even a state of ‘perfection’ in the classical period (cf. the adjective ‘Golden’ above) ought to be re-examined.
But it is in the late period (say, from 300 to 600) that easy narratives run into real trouble. The main objection to splitting our extant texts into neat chronological units and implying that these units represent stages in the history of the language as it was spoken lies in the inadequacy of much late Latin as linguistic evidence. It is true that informal writing tablets have been turning up, quite a few of which are in this selection (19–24, 32–7, 46), and that these do bring out some changes that were in progress, but they are minute in extent compared with the amount of Christian Latin. This was heavily influenced by versions of the Latin Bible, and the Bible was a translation text, full of artificiality and translationese. For example, in much late Latin the infinitive of purpose with verbs of motion is confined to biblical Latin or texts inspired by the Bible (see 39.22, and Adams and Vincent Reference Adams, Vincent, Adams and Vincentforthcoming b). One can learn little or nothing from evidence of this kind about the state of the ordinary language. Even more striking is the use of ad with verbs of saying (with unemphatic pronouns as well as names and nouns) in the Hist. Apoll. (see the detailed discussion, 43 Appendix). This use can be related to the Vulgate version of the Old Testament, but in most respects not to the Romance languages: it looks artificial (and might indeed derive from a lost Greek original). One might it is true argue that in Christian texts a period of Latin is indeed reflected, the ‘Christian period’, but the problem is that translationese and its derivatives may be quite unlike ordinary speech. In the Conclusions to Patrick (40) I have attempted to find a distinction between ‘living Latin’ in Patrick, and biblical Latin. The biblical element is overwhelming, and not much sign of a living language was discovered. There is little point in compiling crude statistics from Christian texts en masse showing the frequency of particular constructions, as if such statistics would point us to patterns of change. Unless biblical quotations and imitations are excluded from the figures (an impossible undertaking), then we would be documenting to a considerable extent the biblical influence on late Latin writing, not developments in speech. Nor is it convincing to assert that Christian writers were addressing the ordinary people and therefore made a habit of using ordinary Latin. The one text, the Latin Bible (in its various versions), that was certainly intended for the ordinary people is replete with translationese, as was remarked above. Many other Christian texts are theoretical and argumentative, and intended for other theorists.
When we move on to a very late period things become different. We find for example a clear awareness of a distinction between that which was ‘correct’ (i.e. classicising), and that which was deviant or modern, and that awareness leads to the appearance of various texts in more than one version (see 42, 43, 44, 48). Those who were in control of what might be deemed classicising Latin can tell us little about the state of the language, but on the other hand ‘deviant’ versions may be very revealing.
As a chronological unit the most coherent and distinctive group of texts in this anthology are the last five (46–50), from the early medieval period. These have a profusion of proto-Romance features, which I summarise here.
Two texts have a regular definite article (49.7, 50.24), and in both the article may combine with de to form a partitive article (49.14, 50.29–30). In two suus is used non-reflexively of a single possessor, equivalent to CL eius, which however has not been completely displaced in one of the two texts (47.15, 49.3). There are some distinctive lexical items, anticipatory of areas of Romance (see e.g. 46.7–8 on fuimus/uado in a Spanish text and 47.7 on ambulo/uado in a Gallic text, 48.2.2 on firmitas in a Gallic text, 50.35, 50.40, 50.55, 50.62–3, 50.65 on teneo, gluto, agrumen, frixoria and smaltire in an Italian text). A Spanish text has a name with a Gothic suffix that was influential in Spain (46.4). There are also signs in these texts of proto-Romance developments in the use of case and prepositions. These are summarised below, 4.4.
I conclude that continuities, real or imagined, between early Latin and later periods need further investigation, that the possible standardisation movement in the classical period is a topic ripe for a detailed study, that statistical studies of linguistic phenomena in texts from the Christian period should distinguish between biblical quotation and imitation on the one hand and free composition on the other, and that early medieval texts offer a huge field for further linguistic study. Any student of the history of a language must have a cut-off point, but a date of AD 600 or thereabouts, a terminus that I (in the past) and others have adopted, has the consequence that some texts in which there is a surfacing of Romance phenomena are excluded.
2 Latin and Greek
For much of the period covered by this volume Latin speakers were in contact with Greeks, and a recurrent theme has been the impact of that contact on Latin.
In the earliest period we find loan-words that had undergone modifications to their Greek base forms, which show that they were acoustic borrowings derived from spoken contacts and not from the reading of high Greek literature (see 2.25 on elephantus, a borrowing of the Greek genitive form, and also the examples collected at 16.2). Sometimes the (non-literary) source may be identified, as for example the Greek of Magna Graecia (see 2.24 on epityra, 10.5.1 on colaphus) or contemporary koine (see 2.5 on machaera). Somewhat later, in Petronius, saplutus seems to have been current in the west, certainly in Gaul (18.10), and in the same work lupatria, if it is a hybrid, is obviously another popular borrowing. Some Greek terms in Latin are all but unattested in extant Greek, a possible sign of their popular character (see 16.2 on pathicus and various other terms, along with remarks about alternative reasons why such terms might have been unattested in Greek). The Latin of Hermeros in Petronius is full of Greek, not all of it paralleled or possible to explain (see also 18.8, 18.17, 18.18, 18 Conclusions). Here is exemplified another source of Greek in the Latin of the early Empire: Greeks in Italy imported Greek into their adoptive language.
But not all of the Greek influence on Latin was exerted at a popular level. See, for the early period, 1.15 on the present participle with accusative object, and 11.16 on the construction est + infinitive. In later Latin Greek was influential via the Bible: see 39.22 on the revival of the infinitive of purpose with verbs of motion, and 43 Appendix, ‘Some remarks about the NT’, on ad for the dative of the indirect object with verbs of saying.
In the Empire in the east Greek and Latin were in particularly close contact in military circles. That is reflected in the passing of epistolary formulae in letters between soldiers from one language to the other. On a Latin formula apparently transferred to Greek see 13.16. For Greek formulae translated into Latin see 20.3, 26.2. In a pair of letters by the same person, one in Greek and one in Latin, we saw the same phraseology in both (23.8–9); on ‘convergence’ of formulae see 20 Conclusions. In Latin letters written in the east Greek loan-words were adopted freely (see 20.3, 22.29, 26.13, 26.14).
The translating of a formula of one language into the other might be classified as ‘imitation’. On the other hand when in such a letter a Greek construction turns up in Latin form, the term ‘interference’ is more appropriate (see 22.21–2).
In other parts of the Empire informal writing tablets have elements from vernacular languages, notably Germanic (see 35.10–11, 36.4) and Celtic (36.8, 37.6–21).
3 Regional Latin
It was noted in the section on periodisation above that in the early medieval texts included here there are proto-Romance features, including regionalisms, mainly of a lexical kind. The regular articloid use of ille in the description of the Basilica of Saint-Denis (49) is not confined in its Romance survival to France, but it is in that text an anticipation of a Romance usage of the area in which the text was written.
In earlier texts we have noted a few usages that were probably typical of the regions from which the texts originate. There are several probable or possible ‘Pompeian’ usages in the document of Eunus (see 15.8, 15.13, 15.14, 15.19–20). The curse tablet from Leicester in its constant omission of intervocalic g provides overwhelming evidence for lenition in the area from which its writer came, whether Britain or Gaul (see 37.2–3). The same text (37.1), and another British curse tablet, also have an odd u-glide in hiatus that can be associated with Britain.
4 Some distributional patterns
Some usages have come up quite often in texts covering a long period, and it may be useful here to select a few of these and to describe any patterns that might have emerged.
4.1 Accusative + infinitive and substitute constructions
In the early second century, in Terentianus, direct quotation is favoured for the reporting of speech (22.10), and centuries later in the Vita sanctae Euphrosynae direct quotation is overwhelmingly preferred (see 47.12). There are late texts in this selection in which the acc. + inf. hardly occurs (see 39.18, 47.12; see also 31.7), and in which, if it does, its use may be marked by ineptitudes suggesting that it was no longer genuinely current (see 47.12). A statistical survey of the late period would no doubt show a decline compared with quod-/quia-clauses (note, however, that in the Passio Perpetuae of the early third century the acc. + inf. is still the norm (see 27.5), even though by that period in early Bible translations the quod-/quia-construction would have been in use on the Greek model). What is interesting, however, in some of our texts is that the acc. + inf. could be treated as stylistically marked and exploited for special effects. This is a phenomenon that is apparent in three extracts. Augustine, who after his conversion made quite frequent use of clauses of the quod-type, in his reply to Publicola sticks rigidly to the acc. + inf., whereas Publicola had used mainly the quod-construction in his own letter (see 31.7). Augustine was distancing himself from his correspondent, who had almost certainly irritated him, and that distancing takes the form of adopting a ‘superior’, classical style. The author of the Hist. Apoll. also seems to have given the acc. + inf. higher status, in that he restricts the quod-construction to direct speeches (43.4). There is also some complexity to his use of the acc. + inf. Finally, at a much later period the more ‘correct’ second version of the Annales regni Francorum sometimes replaces a quod-construction of the first version with the acc. + inf. (48.1.2).
The conclusion to be drawn is that linguistic change does not always progress evenly over time. The old does not die out overnight but may remain available for use by revivalists or purists who wish to make a point.
4.2 Asyndeton bimembre
Asyndeton bimembre, whereby two words of the same syntactic status are placed together without a connective, is an interesting phenomenon. It is necessary to exclude from this category pairs of words that could not alternatively have been coordinated by the addition of, say, et. ‘The big bad wolf’ could be rewritten with a connective (‘the wolf which was big and bad’), whereas ‘the disastrous First World War’ could not (*‘the World War that was disastrous and First’). ‘First World War’ forms a unit, which is qualified by a single adjective, ‘disastrous’. I use old terminology in saying of an example such as this that ‘disastrous’ and ‘First’ differ in rank. The distinction is not always made by commentators, who have sometimes classified pairs of differing rank as asyndetic. Asyndeta can have more members than two, but examples with three (especially) or more members are mundane forms of rhetoric, whereas asyndeta bimembria tend to be stylised or formulaic. Asyndeton bimembre occurs in other Indo-European languages, with some specific types, such as pairs of adjectives of which at least one is privative, widespread (see 3.420; also 6.3 and below). I have made a point of commenting on cases of asyndeton bimembre that occur in the anthology, and the distribution of these comments is revealing of the history of the phenomenon in Latin. It declines over time, and tends to be restricted to certain genres or to formulae. I will say something first about its distribution in this volume, and then about types that have been noted here.
Asyndeton bimembre is common in Plautus (though not so in this selection: but see 3.420, and 17.3 for particular types) and also in Cato (see 4.14–15 for a classified collection of examples from the De agricultura). It was however stressed that some types in Cato are stylistically marked. In the genre of agricultural writing there is a clear decline from Cato to the Res rusticae of Varro (see 4 Appendix). Varro regularly, however, uses three terms together without connectives. Another quite early text in which asyndeton bimembre is frequent is the Johns Hopkins defixio (see 6.1, 6.3, 6.7, 6.8, 6.41). Here genre comes into it. Defixiones share features with the languages of law and religion, in both of which asyndeton bimembre had a place (for asyndeton in prayers see 4.14–15 on Cato, and for legal language in curses, 19.5–6, 19 Conclusions), and the usage occurs in a later corpus of curses represented in this selection (from Mainz: 19.6–7, 19.10, 19.10–12).
In the imperial and later Latin included here (apart from the above curses) the phenomenon is rare. There is an alliterative formula (salua sana) in a private letter of the early Empire (13.15), and another in a freedman’s speech in Petronius (sicca sobria: see 18.12). It was also noted that another alliterative asyndetic expression (agere combined with aginare) used by a freedman in Petronius (in a passage not included in this selection) has now turned up in Mainz curse tablets (see 18 Introduction, 19.6–7), which shows that it was a ready-made formula adopted by Petronius for his speaker. In all three of these examples the linked terms are synonymous. A crowd acclamation of the imperial period (saluum lotum) came up in passing (28.17, 28.25–6). In the whole of the rest of the anthology we have had little call to comment on the phenomenon. The African surveyor uses a non-formulaic expression, nudus saucius, in a context in which it was suggested he was trying to write impressively (25.8–9). We were able to cite a possible case (mendacibus periuris; classified as asyndeton by Bieler Reference Bieler1993: 146) from Patrick (40 Conclusions), in a section (‘Patrick’s Latin is severely limited’) in which we argued against the idea that Patrick wrote as he spoke: he had stylistic pretensions. Finally, pairs of imperatives could be used either with a connective or in asyndeton, and this type of asyndeton seems to have lived on until late (see 45.14; also 22.28, 43.9).
In the types of Latin considered here asyndeton bimembre was recessive, occurring in the late period occasionally only in formulae or in contexts in which it was generically determined or in pretentious pieces of writing. It would no doubt be possible to find many more examples in higher-style writings of the period, but this book is about mundane texts, from which it was mainly absent.
I turn now to some of the categories into which examples can be put. An old type (found in other languages too) is that of which the two members have a repeated fore-element, such as a prefix (3.420, 19.6–7, 19.10). The adjectival privative prefix in- (or ἀ- in Greek) is one such element. Pairs of adjectives also sometimes show just one such privative term (4.6). Alliterative terms, which are common, also in a sense have a repeated fore-element (4.14–15, 13.15, 18.12, 19.6–7, 19.10–12). It is not only alliteration that is prominent in asyndeta bimembria; there are other forms of assonance too (4.14–15). Sometimes the second element is an intensive compound of the first (6.7), or alternatively a compound may precede its simplex (6.8). There are pairs of synonyms and opposites (4.6, 4.14–15, 6.41, 18.12). Semantically the second term may be an intensification of the first (17.3; also possibly 25.8 nudus saucius, where the second is stronger than the first). The second, intensive element may correct the first (= ‘A, nay rather B’: see 17.3). The combination may be disjunctive (= ‘A or B’: see 5.35) or adversative (4 Appendix). Finally, the problem of classification raised by some pairs of verbs has been referred to (3.420, 4.3–4), as when the verbs refer to sequential events. More straightforward are pairs of synonymous verbs referring to a single event (see 6.41): these are definitely asyndeta, and stylistically marked.
4.3 Word order
Variations of word order have been noted in this work, determined by such factors as the source of a text, its stylistic level and genre. The last word has not been written on word order, and there is need for a rigorous compilation of statistics from texts over a long period, and also for an attempt to identify the multiplicity of factors that may determine variation.
I start with the orders OV/VO in relative and subordinate clauses. There was a persistent tendency in Latin for the verb in subordinate clauses to gravitate towards the end (see Adams Reference Adams1977a: 69, with bibliography), with the consequence that the order OV predominates, even in texts in which in main clauses VO is more usual. In the Anonymus Valesianus II, for example, a work of the sixth century, VO predominates by 37:26 in main clauses, but in subordinate clauses OV is preferred by 21:6 (see Adams Reference Adams1976a: 136, id. Reference Adams1977a: 69).
But there are some conflicting patterns in our texts. At one extreme stands the last text, that on falcon medicine. In this in subordinate clauses verb-final position is almost absolute, predominating over non-final position by 28:3 (see 50.49). A proper comparison with main clauses cannot be made for this text, because in main clauses verbs are almost all imperatives, usually with the object following, a normal pattern throughout Latin. It would seem possible nevertheless that placement of the verb late in subordinate clauses persisted until this time in northern Italy. Further investigation of other early medieval Latin texts might be useful. In Old French, for example, there is some preservation of the order SOV in subordinate clauses, but that order is not invariable but alternates with SVO in response to a variety of factors (see Buridant Reference Buridant2000: 747–8).
At the other extreme there is a passage from the Vulgate version of one of the Gospels, for which statistics are given at 38.57. In that VO is strongly preferred in subordinate clauses as well as main, and indeed even more markedly in subordinate. The obvious reason for this is that the Vulgate has retained the order of the Greek. The order OV was not especially common in subordinate clauses in Greek (see Adams Reference Adams1977a: 69 n. 7). We did however detect some unease among Bible translators about copying this VO pattern of subordinate clauses from the Greek NT. The translator responsible for the Vetus Latina version e tended to convert the non-Latin pattern qui misit me (found often in the Vulgate, following the Greek) into qui me misit (38.57).
Another text in which there seems not to be a sharp distinction between subordinate and main clauses in this respect is the Hist. Apoll. In this we saw the order VO preferred in both types of clauses (see 43.5), though the sample was not large and the whole text ought to be examined. The Hist. Apoll. is an odd work, showing biblical influence in other respects too (notably in the use of ad in expressions of the type ait ad eum). It is worth noting that in the Latin letters of Terentianus too VO is preferred in subordinate as well as main clauses. Terentianus was bilingual, and he shows exactly the same preference for VO in the two types of clauses in his Greek as in his Latin letters (for details see Adams Reference Adams1977a: 68–9). The pattern seen in his Latin subordinate clauses may well reflect Greek influence.
A text that adds another dimension to the picture is the Passio sanctarum Perpetuae et Felicitatis. There are variations between different sections of this text. In the narrative of Perpetua there is a predictable distinction between object position in main clauses compared with subordinate. In main clauses VO predominates by 37:11, whereas in subordinate clauses OV is preferred by 16:7 (see 27.1). On the other hand in the compiler’s narrative the word order contrasts sharply with that of Perpetua. He prefers OV markedly, but in main clauses as well as subordinate. There seems to have been a choice available. The compiler’s sections are more formal, and one aspect of that formality seems to have been a preference for OV even in main clauses. It is true that no attention has been given here to other possible determinants of object placement, but on the face of it there are two different traditions at work, a classicising one and another showing a rightward orientation within the Latin sentence (except in subordinate clauses).
There are indeed other signs of this orientation in some parts of the work, with the compiler’s sections contrasting with those of the other narratives.
Thus Perpetua herself regularly places the acc. + inf. after the higher verb, whereas the redactor more often than not places it before (see 27.5). In the two narratives of the martyrs, Perpetua and Saturus, postposition of the genitive outnumbers anteposition to an overwhelming extent (see 27.15). In classical Latin its placement is more variable, with the two orders roughly equal in incidence (statistics are given), and no doubt determined by many factors. On the other hand the compiler shows a greater readiness to place genitives before their noun. By contrast the order NG is far preferred to GN in a number of the later texts in this anthology, and was tending to be adopted mechanically (see Index, ‘word order, position of genitive’). For further details of the differences between the word order of Perpetua and Saturus on the one hand, and the compiler on the other, see 27 Conclusions.
The Conclusions to 27 also have a comparison between a passage of the Gospel of Mark in Jerome’s Vulgate version, and the opening sections of Jerome’s own Vita Malchi. The former is overwhelmingly VO (i.e. rightward-looking) in its characteristics, the latter mainly OV, with some artistic variation. Jerome was drawing on two different traditions, one from the Greek source, the other with classical connections.
The word order of two of the very late texts in this selection, Gregory the Great (45) and the Spanish slate tablet (46), is commented on in the Conclusions to the two texts. Again we see contrasting patterns. Gregory’s Latin has OV characteristics, whereas that of the slate tablet is uniformly VO in type (I am referring not only to objects in relation to the verb, but to local complements of verbs, datives of the indirect object in relation to the verb, genitives in relation to their noun and adnominal prepositional phrases in relation to their noun). There are no exceptions to this general pattern in the tablet. Gregory on the other hand was a learned classiciser, and he was hanging on to literary patterns of the past.
I give finally one further example of the old in competition with the new. In relative clauses in Patrick clitic pronouns (notably mihi, used as the basis of a note at 40.4) are regularly placed next to the relative pronoun, even when there are other words present in the clause. This is the old Wackernagel pattern. It is shown that Patrick could not have picked the pattern up from the Vulgate, where it is rare. It is however an order that is still represented in Bath curse tablets and Vindolanda tablets, and it has to be assumed that it lingered on for centuries. Occasionally however Patrick puts the pronoun before the verb in a relative clause, as usual, but other words of the clause after the verb (see 40 Conclusions, ‘Living Latin’). This represents a shift from a Wackernagel placement to a Romance. This coexistence of the old and the new is suggestive of the language in a state of change. In the narratives of the Passio sanctarum Perpetuae et Felicitatis too we saw traces of Wackernagel placement lingering (see 27.2), but the main feature of clitic pronouns there is that they are usually juxtaposed with the verb. In Perpetua’s narrative such pronouns are usually before the verb in subordinate clauses, but after it in main clauses.
Word order was complex in the later period, and no one approach, statistical, pragmatic or generic, could get to the bottom of it. Individual texts vary. The topic is complicated by the competing stylistic traditions in evidence.
4.4 Case
The relationship between case and preposition comes up in our texts, with innovations mainly to be seen in the very late period.
In earlier Latin we noted a few interesting phenomena, in a general sense anticipatory of things to come. For example, Cato had a taste for prepositions in certain temporal expressions, in which an unaccompanied ablative of time would have been usual (see 5.6). There are partitive expressions with ex/de standing as object of a verb already in early Latin (4.3), a usage that was to lead to the Romance partitive article, which, we saw above, appears in two of our very late texts in the form of de in combination with an articloid demonstrative (49.14, 50.29–30). A genre-related prepositional usage found in Vitruvius and other technical writers is that of ab expressing agency with nouns that are either abstract or denote natural forces or the like (11.5). With these in classical Latin the instrumental ablative was preferred. In Cato there is a common use of unde equivalent to ex quo/qua/quibus, which was later to survive in Romance (4.5). On instrumental unde in a very late text see 50.49.
It is in the texts from 46 onwards that change becomes apparent. In the slate tablet (46), for instance, the various oblique case uses of domus (domum, domo, domi) have been replaced by prepositions (ad, de, in) (see 46.4, 46.5–6, 46.6). Ad has encroached on the dative of the indirect object in the Vita sanctae Euphrosyne (47.8), but most notably in the text on falcon medicine, in which there is a proto-Romance distinction between the inflected dative with clitic pronouns and ad with nouns (50.45). This text also has ad where classical Latin uses the sympathetic dative (50.45), and ad with verbs normally governing the dative (50.66–7). There are too examples of de with a genitival appearance (note e.g. 50.59, where such a phrase is equivalent to an objective genitive), but interestingly none of these is with nouns denoting animates; the genitive inflection is maintained with animates (50.40–1). Similarly in the description of the Basilica of Saint-Denis the genitive remains in use in names (49.1), and in the Visigothic slate tablet (46) there are six or seven genitives, all of them in personal names. Finally, the text on falcons has not a single instance of the instrumental ablative: its functions have been taken over by de and cum (see 50.33); for notes on uses of de in late texts see also 49.15, 49.18, 50.29, 50.29–30, 50.30–1, 50.31.
4.5 Relative clauses
Attention has been drawn in the commentaries to two types of relatives clauses in particular, the relative-correlative type, and attractio inversa (for examples see e.g. above, 1.11). In the first the relative clause is preposed, and picked up in the main clause by a resumptive element (e.g. Cic. De orat. 2.248 quoscunque locos attingam, … ex eisdem locis fere etiam grauis sententias posse duci). In the second the relative clause again precedes the main clause, but is itself preceded by an antecedent, which however takes its case from the relative clause not the main clause (e.g. Ennius, Euhemerus III tum Saturno filius qui primus natus est, eum necauerunt; filium might have been expected). The second type often seems closely related to the first, in that the antecedent looks to have been fronted out of the relative clause of a relative-correlative construction.
The distribution of these constructions is discussed by Probert and Dickey (Reference Probert, Dickey, Adams and Vincentforthcoming), and Halla-aho (Reference Halla-Aho, Adams and Vincentforthcoming). I comment here on their distribution in the texts collected in this volume.
Both constructions are common in early texts (for relative-correlative see 1.11, 1 Appendix 2, 4.6–7, 4.10, 4.11; for attractio inversa 1.11, 1 Appendix 2, 4.10, 4.11; these examples are in Ennius and Cato, but in the notes other republican attestations are cited as well). Thereafter both all but disappear (in our material, it must be stressed), except in some texts of distinctive type. In the surveyor’s inscription and the letter of Publicola there are unremarkable relative-correlatives of imperfect type, in that a resumptive element is lacking (25.22, 31.3). Apart from that there are examples of both types, first, in curse tablets (relative-correlative 34.3–4, 34.6, 36.5–6; attractio inversa 33.12–15), and, second, in the theological writer Patrick and other Christian texts, including the Vulgate (see 39.10). Curse tablets are traditional in language. Christian examples are derived from the Greek Bible. The relative-correlative type seems to have been suited there to gnomic generalisations, as found for example in the Sermon on the Mount.
5 Some remarks on genre
5.1 Technical writing
Technical texts from the early Republic (5 Cato) through the classical period to late antiquity are included in this anthology (11 Vitruvius, 25 surveyor’s inscription, 29–30 Pelagonius, 42 Physica Plinii, 50 text on falcon medicine). Such works have traditionally been regarded as manifestations of sermo plebeius (a term favoured by Cooper Reference Cooper1895). It is impossible to characterise them in a simple formula, and sermo plebeius will not do.
There are to be sure departures from expected norms in some of these texts. The surveyor for example has the indicative in consecutive clauses (25.16), and Vitruvius has some nouns in unusual genders (11.1). Pelagonius uses coarse excretory and anatomical terminology that would not have been admitted in high literature (30). The Physica Plinii survives in more than one redaction, with an earlier version being ‘improved’ later, and that suggests that the Latinity of the earlier one had not entirely met with approval.
But technical Latin cannot simply be equated with non-standard. Writers on technical subjects had varying control of educated norms, ranging at one extreme from that of the elegant classicising writer Celsus to that of some of the crude medical texts of late antiquity. The manifest stylistic aspirations of some of the above writers, for example the surveyor and Pelagonius, have been pointed out (e.g. 25.8–9, 25.10–11, 29.1; see too the Conclusions to 25 and 30). Technical Latin should be treated as what it is, technical, and its particular features described. Technical terminology in all the above texts has been discussed in the commentaries. Categories of word formation have been noted, such as the substantivising of adjectives and participles in the neuter (see 5.2, 5.11, 11.13) (see also 11.4, 11.12 on suffixal derivatives in Vitruvius). There are also features of syntax, which, though not unique to technical writing, may be distinctive of such genres. For example, in Cato’s medical chapter paratactic conditional sentences are an obtrusive element (see 5.6), and these we paralleled in a parody of doctors’ talk found in Plautus (see 5.6, 5.15). Various turns of phrase found in this passage of Cato continued in medical writing for centuries (see 5.1, 5.2, 5.2–4, 5.5, 5.12). The most striking feature of syntax in technical texts noted here is the constant use of ab with passive verbs expressing the agency of some natural force in determining an event (see 11.5 on Vitruvius compared with Cicero, and on a few other technical writers).
On the apparent ‘non-standard’ elements in Vitruvius, see particularly 11 Conclusions, ‘Other divergences from classical practice’, where it is pointed out that various departures in Vitruvius from the usage of Cicero and Caesar were shared by him with a wide range of educated writers, such as some of Cicero’s correspondents, Varro, Livy, Nepos and Virgil. What this shows is not that these writers were using sermo plebeius, but that the educated norms of the period should not be constructed from a pair of writers.
5.2 Non-literary documents and some genres of Latin literature
Writing tablets have been exemplifying usages in certain genres of Latin, particularly the shorter poems of Catullus and Horace’s Satires, and making it clear that lower genres drew on genuine colloquial varieties. See 13.10 (Catullus and Horace and intensive compounds with dis-), 21.6 (Catullus, Horace and iucundus), 33.3 (Catullus and a use of inuolo). The Pompeian graffito also has three usages found in Catullus, ausculto, pathicus and uerpa (see 16.2, 16.3 and also 16 Conclusions). The combining of ago with the obscure verb agino is now attested in Mainz curse tablets of the first century AD, a pairing previously only found in a freedman’s speech in Petronius (19.6–7), of much the same date. Petronius was (at least up to a point) drawing on real speech in creating the speech of his characters. The future participle factura accompanied by a conditional in the Vindolanda letter adds a dimension to the history of this construction (21.7). See also 14.7 (on pusillus in a non-literary Augustan letter, and Catullus and Horace). Finally, I mention in passing that the fragmentary letters of Augustus (which however are literary in attestation, not documentary) also have parallels with e.g. Horace (see 12 Conclusions).
6 ‘Informal’ texts and Latin
I take up finally a term that is used in the title of the anthology. ‘Informal’, like all stylistic designations, is a vague term, and it might be useful here to define some of the categories of texts in the anthology for which it seems particularly appropriate.
Private letters on mundane subjects that were not intended for publication are usually devoid of literary pretension. Various private letters on writing tablets are included here (12–14, 20–4, 26; I include 12, by Augustus, which was not published but is quoted by Suetonius), just one of which (24) is exceptional in its use of studied language. Caelius’ letter (9), along with others by him that are extant, survives in literary manuscripts and does have a more serious content, but his extant corpus has striking departures from literary norms. In another letter of a literary kind (17), Seneca quotes his own irritable speech directed at two slaves, as well as their replies, and there are several usages in these spoken parts that are virtually excluded from literary texts.
The Magerius mosaic (28) records verbatim shouts of the crowd at a show in the arena.
Educated writers even working within a formal literary genre may, as Seneca did (see above), deliberately present mundane speech or conversation. Plautus (2, 3), as a playwright, had to do so by definition, though his Latin has a huge stylistic range. The classic depiction of trivial speech is that by Petronius (18). The passages from the Rhet. Her. (7, 8) are meant to illustrate run-of-the-mill style, and they also contain speech. The jokes collected here (10) were delivered informally, in spoken form. The Hist. Apoll. (43) has speeches as well as narrative, and there is at least one syntactic distinction between the narrative and spoken parts (43.4). Gregory the Great (45) puts spoken Latin into the mouths of humble characters, admitting usages not usual in his normal discourse. The Visigothic slate tablet (46), though it is strictly a legal document, records conversation, which even has a regional Spanish feature.
Finally, the Vita sanctae Euphrosynae (47) has spoken parts.
A distinction of a specific type may be made between (e.g.) the letter of Seneca referred to above, and most of the writing tablets in this volume, not only the letters just listed, but also the curse tablets (6, 19, 32–7) and legal tablets (15, 41, 46). Seneca was a prominent literary figure, who just happened to reveal in the letter how he might speak to (or at) slaves in anger. His outburst was certainly not ‘formal’. On the other hand those composing (e.g.) curses were working within a genre that might be described as formal, in that gods were addressed and formulaic language was to some extent used. But writers of curses were usually poorly educated, and their efforts are full, for example, of spelling errors. Defixiones might be ‘formal’ texts in a sense, but in another sense they have Latin that would not be thought by the educated suited to formal contexts. Amid their formulae, curses have usages redolent of speech (such as the combining of ago with agino: see above). We thus have the educated sometimes writing down, and the uneducated writing up.
A number of the texts in this volume, as we just saw, fall into the category ‘technical’. Technical Latin is difficult to characterise, and it has been given its own section above. There are again informal features in such texts.
The remaining texts include a narrative deliberately written (up to a point) in a mundane style (1), an obscene graffito (16), the (formal) letter of Publicola that was indirectly portrayed by Augustine as inappropriate in style (31), and the various texts with more than one version (42, 43, 44, 48). In this last group one of the versions is in effect treated by a later redactor as stylistically defective.
The features of all the extracts mentioned in this section have been summarised in the Conclusions to each commentary. Recurrent usages include types of parataxis (3 Conclusions; henceforth in this paragraph ‘Conclusions’ should be added to all single-digit numbers given; 7.8–9, 15.19–20, 18.5, 18.10, 18.22, 18), certain reinforced demonstratives and adverbs (3, 6, 8, 9), hanging nouns (2.25, 3.416, 4), low-register lexical items (5, 6, 10, 12.3, 12.4, 18.8, 22.13, 27 Conclusions, ‘Perpetua’s Latin’), incoherence or carelessness of expression (5, 11.31, 22.29, 25.28), intensive compounds in dis- (6, 9, 13.10), popular Greek borrowings and Greek words unattested in Greek (2.5, 2.25, 16.2, 18), deviant morphology and syntax (as -rus and -aes, 6, 13, 14, per + nominative 15.5–6, locative with verb of motion 22.14–15, fossilised tempus 22.17, ablative gerund exiendo 22.31, cum + accusative 22.22, adiuto + dative 6.29, 22.28, consecutive ut followed by indicative 25.16).
