In VI–XI six types of asyndeton have been examined that are marked by a grammatical (usually morphological) feature of both or one of the terms. The diversity of the types, and particularly the variations in their distributions, are already enough to make it clear that there is no point in trying to come up with a single adjective or phrase that might characterise asyndeton bimembre (‘sacral’, ‘tragic’, ‘reflecting the ancient carmen’, ‘expressing rapidity’, etc). The different types vary particularly in their frequency across genres, literary and non-literary.

The term ‘list’ is not particularly helpful in characterising asyndeton and its types, though it (or an equivalent) does have a place in the literature. One ‘special type’ of asyndeton, according to Hofmann and Szantyr (1965: 830), is ‘asyndeton enumerativum’. But virtually any asyndetic sequence, particularly of names, nouns or adjectives, may be seen as a list. A series of adjectives describing the physical features or character of a person is a list of personal characteristics, the itinerary of a traveller is a list of places, the names of persons appointed to some organisation for a fixed period make up a list of officials or officers, and so on. When is an asyndetic sequence not a list? A pair or series of verbs describing actions in a temporal sequence (they stopped, turned, fled) would not be well described as a list. Pairs of terms with certain semantic relationships, such as opposites of various sorts (left/right; up/down; good/bad, go/return), are not list-like. A pair of verbs consisting of a simplex followed by its compound is not a list. It may be more helpful to identify types of lists. One of these, of some importance in relation to asyndeton, I will call ‘open-ended’ or composed of ‘illustrations’. I start however with a different type, which is perhaps more familiar to anyone using the term ‘list’. This type expresses a totality or finite set.

Pairs of imperatives, used asyndetically, are common from early Latin (see e.g. Adams 2016: 558 listing examples from the early Republic; see too XXII.4.4 for asyndetic pairs of imperatives in Umbrian). The significance of the structure (in relation to overt coordination) in the early period may be illustrated with some details about one of its manifestations, that comprising i (or ite) + imperative (see particularly the discussion of Sjögren 1900: 82–91). In Plautus (and also Terence) this type of asyndeton is commonplace, and it tends to occur in formulae or idioms suggestive of everyday speech. By contrast in Cato Agr. coordinated pairs of imperatives (with -que the usual connective) far outnumber asyndetic (by about 80:5: see Adams 2016: 80), but a factor is that most imperatives in Cato are of the -to form, and it must have been the commonplace present imperative that was usual in ordinary speech. In all examples cited below from Plautus at least one member of the pair is a present imperative.

‘Asyndeton’ has been used by classicists in mixed ways. In this book I use the term as it is used in modern linguistics, to refer to a form of coordination. Various other phenomena, though interesting in their own right and often labelled ‘asyndeta’ by classical commentators and others, are left aside, worthy as they may be of study. I start with ‘coordination’.

If two terms are juxtaposed without a coordinator they need not be in asyndeton, though it is commonly assumed in commentaries that any juxtaposed words of the same part of speech must be asyndetic. There are various relationships between two words that may rule out the insertion of a coordinator. For example, if the two are adjectives, they may differ hierarchically (see e.g. Quirk et al. 1972: 550–1), a distinction sometimes referred to as ‘rank’. Or one may be attributive and the other predicative, or a secondary predicate (see below). A pair may form an ‘appositional compound’. Again, one term may be a modifier of the other, with overt coordination possibly unnecessary or inappropriate, though this type is rather nebulous. All these categories will be illustrated below. Perhaps most important are the roles of adjectives, such as attributive versus predicative. The distinction is sometimes neglected in literary commentaries.

In republican laws asyndetic coordination is widespread, whereas in early prayers in Latin, despite assumptions to the contrary (see e.g. Calboli 1997a: 800), it is uncommon. Umbrian religious language is a different matter: in the Iguvine tablets asyndeton is the normal type of coordination for pairs (see below, 4 for a detailed account of Umbrian).

Asyndeton bimembre was old (see e.g. III.1, XIX.1), but I have resisted the idea that it was ‘undoubtedly older’ than explicit coordination of pairs (III.1). -Que and its root were ancient too, and I would prefer with e.g. Dunkel (1982) to leave open the matter of relative antiquity and to assume a long coexistence (III.1). From the very beginning in Latin asyndeton bimembre was diverse in its types and in their stylistic level. It has however been exposed to snap judgements derived from a failure to look at the distribution of its forms in a range of genres. For example, at XXIV.5.1.1, 8 (cf. 5.1.1, 1) Skutsch is quoted as saying that two adjectives with one noun are characteristic of ritual language. Asyndetic pairs of adjectives are common in many writers, some of them working in mundane genres (see below, this section), and indeed one type (consisting of pairs of judgemental adjectives in open-ended lists placed at the end of cola) we have related speculatively to a pattern of speech (see IV.4; cf. V.2, and also below, this section). Or again, Ogilvie (1965: 730), commenting on Livy 35, describes a pair of privatives (inuisitato inaudito) as ‘almost sacral’. Privatives in asyndeton, whether two together or one juxtaposed with a different type of adjective, are so widespread in a variety of genres, e.g. oratory, historiography, Horace’s Satires, that a whole chapter has been devoted to them (VI). Similarly Jocelyn (1967: 175) on Enn. trag. 9 pugnant proeliant refers to the ‘official language’, but while some pairs of verbs belonged to legal language this is not one of them, and asyndetic pairs of verbs fall into diverse categories (see below, this section).

Part 3 on semantic types is highly selective, dealing as it does only with opposites, itself however a big subject, and with terms asyndetically specifying different family relationships. This second category has some overlap with that dealt with in XI. I have been selective in this part because semantic types are commented on constantly in the chapters on literary texts. I give a few cross references below.

Asyndeton bimembre is common in Plautus but not overwhelmingly so. There is little point in trying to produce definitive figures, because there are so many variations on the basic (single-word) type, such as word + short phrase, word + long phrase, short phrase + word etc., not to mention unitary pairs within longer asyndetic sequences, which may be treated either as asyndeta bimembria or simply as components of long asyndeta. Phrasal pairs are also common. I concentrate in this chapter mainly on the simple type, or types with only minor modifications of that.

In this chapter the different genres of Horace’s work, the Satires, Epistles, Odes, Epodes and Ars poetica, are kept apart. Striking generic differences in the use of asyndeton will emerge, between on the one hand the Satires and first book of the Epistles, and on the other the Odes (and Epodes).

Historiography I have split into two chapters, putting Livy in a separate chapter. Sallust and Tacitus are kept together because of the commonplace view that Tacitus was influenced by Sallust, and the pair will be compared at the end of the chapter (6). A brief treatment of the remains of the early annalists precedes that of Sallust.

Types of living beings are sometimes expressed by pairing the masculine term for that being with the feminine. For example, the totality of divine beings may be rendered by ‘(all) gods (and) goddesses’. Such pairings in Latin show variations between syndetic and asyndetic coordination, with the asyndetic variant common in legal language but coordination usually preferred otherwise. I have just used the term ‘totality’, but it is misleading without specification. ‘Men (and) women’ may refer to the infinite number of male and female adults in the world, but the phrase is more likely in ordinary narrative to denote a finite, even small, group in a particular context. The term ‘merism’ is all very well in idealised accounts of Indo-European poetics, but most people in the real world do not speak only in universals. Watkins’ definition (1995: 9, 15) of merism is ‘a two-part figure which makes reference to the totality of a single higher concept’. West (2007: 99–100), more clearly, refers to ‘pairings of contrasted terms, as an emphatic expression of the totality that they make up’. The method of coordination in male–female pairs in Latin seems unaffected by the difference between a finite set and an infinite.

A contrived structure with asyndeton is that in which an asyndetic pair is followed by another pair, asyndetic or syndetic, the members of which refer back to the members of the first pair. Usually the first member of the initial asyndetic pair is picked up by the first member of the following pair (and the second member by the second), but there is sometimes chiasmus, such that the second two terms allude to the first two in reverse order.