Varro's De lingua Latina (= Ling.) is a treasure trove of information.Footnote 1 Of the originally twenty-five books, six have come down to us more or less complete. Among these, Books 5–7 give us many hundreds of etymologies, and Books 8–10 discuss the question whether Latin morphology is regular or not. What Varro rarely comments on is sociolinguistic variation. The sociolinguistic comments in Varro's work can almost be counted on one hand. For instance, in 5.162 Varro remarks that cenaculum, from cena ‘dinner’, means ‘attic’ in Roman Latin, but that the original meaning was, as one might expect, ‘dining-room’, a meaning preserved in the non-Roman dialects of Latium, Falerii and Corduba in Spain; the meaning also lives on in religious uses in Lanuvium. In 7.96, Varro tells us that some words are pronounced with -ae- by some, but with -ē- by others. We know that non-Roman varieties of Latin monophthongized -ae- earlier than the Roman dialect did. Varro mentions pairs like scaena / scēna ‘stage’, a loan from Greek σκηνή, which shows that -ae- is hypercorrect here. Interestingly, only in one such pair is the variant with -ē- ascribed to country people, in the name Maesius / Mēsius, and this is indeed the only pair where the diphthong -ae- is original rather than the result of hypercorrection.
The passage I want to look at in this short note is Ling. 6.59; here, Varro comments that nouissimus, the superlative of nouus ‘new’, was avoided by his teacher Aelius Stilo and by other men of Aelius’ generation. The textual transmission of this chapter is largely unproblematic, and all editors and translators have treated its interpretation as equally straightforward. However, Varro's phrasing is ambiguous immediately after he says that nouissimus was avoided, and this means that it is by no means obvious why this superlative was avoided. We will begin by looking at the transmission.
1. THE TEXT OF LING. 6.59
The text of the De lingua Latina is transmitted on a parchment codex written in Monte Cassino in the eleventh century, but now kept in Florence in the Biblioteca Medicea Laurenziana. This manuscript, shelf-mark LI.10 and referred to as F (for Florentinus), is in the Beneventan script. For our passage, F is not available, but the relevant folios still existed in 1521, when Petrus Victorius and Iacobus Diacetus compared their copy of Pomponius Laetus’ edition with our manuscript. They noted all divergences, and these are marked Fv in critical apparatuses. Agreement between F and their copy of Laetus’ edition is indicated as (Fv).Footnote 2 For our chapter, we also have an element of indirect transmission: Aulus Gellius dedicates an entire chapter to Cicero's avoidance of nouissimus (10.21) and quotes Varro.
This is what the text of Ling. 6.59 looks like in my recently published edition of Varro (see n. 2):
a quo etiam ‘extremum’ nouissimum quoque dici coeptum uolgo, quod mea memoria ut Aelius sic senes aliquot, nimium nouum uerbum quod esset, uitabant; cuius origo, ut a uetere uetust<i>us ac ueterrimum, sic ab nouo declinatum <nouius et> nouissimum, quod ‘extremum’.
dici Antonius Augustinus (secutus Gellium): dico Fv uetust<i>us Bentinus (secutus Gellium)ueterrimum Bentinus (secutus Gellium): ueterrimus (Fv) nouius et add. Antonius Augustinus (secutus Gellium)
As can be seen, only minor changes have to be made, all based on Gellius’ text, and these were already made by early editors such as Antonius Augustinus or Bentinus. The passive infinitive dici is required by the syntax. uetustius rather than uetustus is needed because we expect a comparative between the base form and the superlative. ueterrimum rather than ueterrimus is required to keep the same gender throughout the series of positive, comparative and superlative. Finally, the comparative nouius must have dropped out between the positive and the superlative. The text is thus quite certain. This is how I translate it, leaving the difficult part in Latin:
From this also extremum ‘last’ popularly began to be called nouissimum ‘newest’, which in my time both Aelius and some older men avoided nimium nouum uerbum quod esset. Its origin is this: just as from uetus ‘old’ we get uetustius ‘older’ and ueterrimum ‘oldest’, so from nouum ‘new’ was derived nouius ‘newer’ and nouissimum ‘newest’, which is extremum ‘last’.
2. THE PROBLEM
The question now arises how we should interpret nimium nouum uerbum quod esset. Kent translates, or rather paraphrases, as ‘on the ground that the proper form of the superlative of this word was nimium nouum’;Footnote 3 in other words, the objection Aelius Stilo and other men of his generation had was that nouus did not have a synthetic superlative so that an analytic one had to be formed instead. However, nimium is not normally used to form such analytic superlatives.
On the other hand, Riganti translates as ‘ritenendola una parola troppo recente’, ‘considering it too recent a word’.Footnote 4 Thus, according to Riganti, the objection by Aelius Stilo was not that the correct superlative should be nimium nouum but that there should be no superlative at all for nouus. Flobert is in broad agreement.Footnote 5 He believes that Aelius’ objection was that nouus is an ‘absolute’ adjective, unlike recens ‘new’. One thing may be newer than another, and in such cases a comparative recentior or a superlative recentissimus may be used. But nouus, as an absolute adjective, means ‘completely new’ in and of itself, so no gradation is possible. Aelius, then, objected to nouus losing its absolute status.
On neither side do we even find an acknowledgment that there is a problem. In fact, none of the three scholars seems to have noticed that the passage can be read in more than one way. But which way is correct?
3. ADJECTIVE GRADATION
Before we can answer this question, it may be worth looking at the gradation of adjectives in Latin in general and in Varro in particular.
In English, we find three types of gradation: suppletion (good, better, best), synthetic gradation (warm, warmer, warmest) and analytic gradation (beautiful, more beautiful, most beautiful). Suppletion is restricted to a handful of high-frequency adjectives learnt in early childhood. Synthetic gradation is used for short adjectives, and analytic gradation for longer ones; originally, synthetic gradation was used for native Anglo-Saxon vocabulary and analytic gradation for words from Norman French, but since Anglo-Saxon adjectives tended to be shorter, this old pattern was reanalysed.
Latin is very similar to English. Suppletion is restricted to high-frequency adjectives like bonus ‘good’ and malus ‘bad’. Synthetic gradation is used for most other adjectives, but analytic formations can be used for adjectives where a synthetic comparative or superlative would result in too many vowels following each other (idoneus ‘suitable’, magis idoneus ‘more suitable’ rather than *idoneior;Footnote 6aiio ‘I say’ and similar words are phonologically acceptable, since -ii- here stands for a geminate semivowel).
Varro is very much aware of suppletion and discusses it in Ling. 8.75–6, without offering a solution to what he considers an anomaly in the system. The distinction between synthetic and analytic gradation does not come up in his work, because it does not affect the regularity of inflection.
Varro is also aware of absolute adjectives. One of his examples is luscus ‘one-eyed’, which he points out cannot have gradation because a man may be one-eyed or not, but no man is more one-eyed than another (9.72).
All this is very sensible, but there are the occasional oddities in the argument, and these are not restricted to Book 8, where Varro plays devil's advocate and argues that inflection is by and large irregular. We will briefly look at some of these oddities.
Let us begin with 8.77. This chapter is not so much an oddity as an indication that Varro, despite being a polymath, did not know everything. He argues, deliberately pig-headedly, as we are in the midst of Book 8, that based on the superlative pauperrimus ‘poorest’ we would expect the positive of this adjective to belong to the second / first declension, while in reality it belongs to the third. And indeed pauper does belong to the third declension in Varro's time, but he overlooks the fact that in early Latin we find attestations of the second / first declension (paupera mulier in Plaut. fr. 195 / inc. fr. 66). pauper became an adjective of the third declension only by association with its third-declension antonym diues ‘rich’.
In Ling. 8.76, Varro claims that caesius ‘grey-eyed’ has a superlative caesissimus but no comparative *caesior. In actual texts, neither the comparative nor the superlative is attested. Varro's intuition may, however, be correct. The adjective is not frequent to begin with, so the lack of attestations may not mean much. The adjective ends in -ius, which makes synthetic gradation more difficult; a comparative in *-iior is generally avoided. The expected superlative should be *caesiissimus or *maxime caesius. However, the absence of a comparative, synthetic or analytic, has little to do with morphology; ‘grey-eyed’ is an absolute adjective, so there are semantic objections to a comparative. The superlative, on this account, does not indicate the person with the greyest eyes but someone in whom they are a particularly noticeable feature.
We can now turn to another part of Ling. 8.77. Here Varro discusses how many syllables are added in various comparatives and superlatives when compared with the positive forms. The text is slightly problematic. The context demands that we should restore a comparative sacrior ‘holier’ with Dahlmann; this restoration is practically certain, because Varro discusses the number of syllables that the comparatives and superlatives of tener ‘tender’, macer ‘lean’ and sacer ‘holy’ have. The trouble with this is that sacer does not have a comparative in real texts. A superlative sacerrimus is found, but sacrior does not exist. This is not an accidental gap either; the adjective is not rare, and we know why there is no comparative: sanctus ‘holy’ is a near-synonym, and wherever a comparative of sacer is needed, sanctior does duty for it. Why does Varro claim that a form exists when it really does not? The answer is probably that native-speaker intuitions are not always reliable, and that we all get things wrong after a string of grammaticality judgements that leave us tired and confused.Footnote 7 A similar error of judgement occurs in Cicero: he condemns the regular superlative of pius ‘devout’, piissimus, in Phil. 13.43; the form one is supposed to use is pientissimus. However, Caper (Cic. Epist. fr. 18[17].3) informs us that Cicero did in fact also use piissimus.
The biggest oddity in the discussion of gradation is found in Ling. 9.72. Why do the adverbials mane ‘in the morning’ and uespere / uesperi ‘in the evening’ not have comparative and superlative forms? Varro does not give the obvious answer, namely that these are fossilized ablatives in -ĕ (and -ī is a fossilized locative). Plautus still has an ablative mani (Mil. 503), and uesper never died out. Adverbs that are derived from adjectives end in -ē and follow the normal gradation patterns. Varro's answer to the problem is based on logic rather than morphology: on a time axis, there can be earlier or later, but not more or less.
This purely logical explanation leads to a conundrum. Why do we then say magis mane ‘more in the morning, earlier in the morning’ and magis uespere ‘more in the evening, later in the evening’? And why does ‘more’ mean ‘earlier’ when it comes to the morning, but ‘later’ when it comes to the evening? Varro does not have a good explanation. Modern linguists do believe in the importance of logic, hence the ubiquitous courses on truth-conditional semantics in most departments of linguistics, but they do not consider it the only principle of semantics. Morning and evening are, as it were, extreme times. If magis ‘more’ is added, it makes these times more extreme, more removed from our reference time, broad daylight. An early morning is more extreme than a late morning, just as a late evening is more extreme than an early one.
We can now return to Ling. 6.59.
4. A SOLUTION TO THE PROBLEM
We have seen how Varro discusses the gradation of adjectives (and adverbs). There are, admittedly, some oddities, but Varro is aware of the distinction between suppletion, synthetic gradation and analytic gradation (magis mane). He is equally aware of absolute adjectives. And while he does not yet have the terminology for all these concepts, or all the answers to the problems that we have today, he nevertheless manages to express the issues clearly in Books 8 and 9.
Why, then, does he not tell us clearly and unambiguously why Aelius Stilo avoided nouissimus? Was it because we need an analytic superlative, or because nouus is an absolute adjective?
I believe that this is actually the wrong set of answers to choose from. The discussion of the gradation of adjectives is restricted to the morphological part of the work, to Books 8–10. It is not an issue that bothers Varro in the etymological books, where he does not care about the distinction between synthetic or analytic, or about absolute adjectives as opposed to gradable ones. Two other issues matter to Varro in the etymological books (Ling. 5.2): the formal derivation of one word from another (ἐτυμολογία), and the meaning of words (περὶ σημαινομένων). The formal derivation of nouus is given in Ling. 6.58, where Varro believes it to be a loan from Greek νέος (today we would say that the words are cognates). Ling. 6.59 then discusses the meaning of nouus, and here we learn that it means extremus ‘last’ (as in nouissimum agmen ‘the last part of a military column’). This meaning is not fully predictable from the base form nouus ‘new’; it is an innovation, and that is what makes nouissimus a nimium nouum uerbum, an ‘overly new word’ that Aelius Stilo did not like.Footnote 8nouum in the objection comes close to ‘unusual, odd’, a meaning we also find in Plaut. Cas. 434 and many other authors.
Kent is wrong in assuming that Aelius Stilo wanted an analytic superlative. The inflectional morphology of nouus is irrelevant here; further proof lies in the fact that in our chapter Varro compares uetus, uetustior, ueterrimus with a suppletive comparative taken from uetustus ‘old’, but does not consider this worth commenting on (the morphologically expected comparative ueterior is attested only once, in Cato, Hist. 21 Peter / 26 Cornell).
In the opposing camp, Riganti is not wrong, but she is vague. Flobert is clearer, but as wrong as Kent, since nouus is not an absolute adjective. nouior was apparently considered unproblematic by Aelius Stilo, who only objected to nouissimus.
Aelius Stilo and old men of his generation objected to nouissimus ‘last’ on the grounds of its unpredictable meaning; they were not the last ones. Krebs and Schmalz show that Varro's younger contemporary Cicero, to whom the De lingua Latina is dedicated, studiously avoided nouissimus, using the word only once in QRosc. 30, a work of his youth in which he admitted other usages that he later shunned as he became more prescriptive.Footnote 9 Caesar, Livy and Tacitus, on the other hand, did not mind nouissimus, and neither did Varro, who has it in Rust. 1.2.11, 1.14.4 and 3.9.18; in our work, he uses it in 9.73, glossing magis uespere ‘later in the evening’ as nouissimum uesper ‘last part of the evening’. Varro respected his teacher, but did not follow him slavishly.
5. CONCLUSION
Varro's teacher Aelius Stilo did not avoid nouissimus on the grounds that we need an analytic superlative, or because nouus should not have a superlative at all. He shunned the word because of its meaning ‘last’, a meaning which is not fully predictable from the base form nouus ‘new’. Our passage should thus be translated as follows:
From this also extremum ‘last’ popularly began to be called nouissimum ‘newest’, which in my time both Aelius and some older men avoided on the grounds that it was an overly new word. Its origin is this: just as from uetus ‘old’ we get uetustius ‘older’ and ueterrimum ‘oldest’, so from nouum ‘new’ was derived nouius ‘newer’ and nouissimum ‘newest’, which is extremum ‘last’.
If we want to be less literal but closer to what Varro intended, we could translate as ‘on the grounds that it was an overly unusual meaning of the word’.
