This thesis treats the topic of hyperbaton, or discontinuous noun phrases, and its use in Cicero’s oratory. It considers the totality of all instances of this linguistic phenomenon – over 2,200 cases – in twenty-one of the orator’s speeches, and thus constitutes the first full-scale, systematic study on the topic. Through careful mapping of the syntactic and semantic features of these instances, as well as of continuous noun phrases that form a basis for comparison, it aims to define the recurring features of this complex construction, and provide some indications as to when it is used.

Chapter 1 provides an introduction to some of the issues surrounding hyperbaton, and questions raised by previous scholarship. This includes the definition of ‘discontinuous noun phrase’, potentially useful theoretical frameworks for understanding this aspect of Latin word order, and some considerations regarding the selection of data and the methods for analysing them. This chapter also introduces my corpus, and a description of how I approached the material.

In Chapter 2, I look at instances of hyperbaton where the intervening word is a preposition, of the type summa cum laude. While previous scholarship has sometimes argued that this word order is unremarkable, I show that the 316 instances in my data allow to determine rather clear criteria for when it is used and when it is not. The word order only occurs with a handful of monosyllabic prepositions. Furthermore, one subtype is characterised by the combination of a determiner and a generic or anaphoric noun. Other cases occur with other types of modifiers, and these seem to be restricted to certain semantic categories, such as adjectives of quantity and size, and to those that are contrasted with another element.

Chapter 3 examines 375 cases where the word that separates the two parts of the noun phrase is a personal or demonstrative pronoun. In nearly all the instances, the pronoun appears in the second position of the clause. Previous scholarship has suggested that this placement depends on prosodic factors, and that the pronouns behave like enclitics. I address this question with regard to the instances in my corpus, and argue that while there are some indications in that direction, in particular similarities with instances treated in Chapter 2, there are also a number of cases where such an analysis is improbable for various reasons.

In Chapter 4, I propose to group several morpho-syntactic classes of interveners that all perform an adverbial (in a broad sense) function in the clause. I maintain that there is a kind of sliding scale of connectedness with the noun phrase within which these words appear, where some should be analysed as part of that noun phrase whereas others are clearly external. Even the latter group, however, usually stand in a semantic or pragmatic relationship with part of the discontinuous noun phrase, and their placement can thus be understood according to a gradual, logical construction of meaning in the clause.

Chapter 5 groups instances of hyperbaton where the intervening word is a mandatory constituent of the clause: these constitute several different categories, and the data discussed in this chapter are therefore heterogeneous. A first section treats intervening pronouns such as quisque and similar, which show a relationship with part of the split noun phrase that is very similar to the instances discussed in the preceding chapter. Similarly, a semantic or pragmatic relationship can very often be supposed when the intervening word is an adjective that functions as attribute to a mandatory constituent of the clause. Moreover, the chapter includes a short section on predicative adjectives appearing within noun phrases, before turning to the largest portion of the data treated in this chapter, which consists of mandatory nouns that split other mandatory arguments. The last section treats cases where a noun, participle, or adjective splits its own complement. Some of the instances can be analysed in a similar way to the data in the preceding chapter, while others are more akin to cases of ‘verbal hyperbaton’, which is the object of Chapter 6.

‘Verbal hyperbaton’ has been proposed to form a separate category of hyperbaton in previous scholarship. In this chapter, I evaluate some of the claims made, comparing this data to that of the entire corpus. I maintain that instances of verbal hyperbaton, except for some specific patterns, are not particularly different from the categories of hyperbaton discussed in the other chapters. Instead, verbal hyperbaton can largely be explained through consideration of the same factors that are significant for the occurrence of hyperbaton at large. I also discuss the potential influence of prose rhythm and other artistic factors.

Chapter 7, lastly, treats cases of hyperbaton where multiple constituents appear within the discontinuous noun phrase: I call these instances ‘long-distance hyperbaton’. These complex constructions display many of the features revealed in the preceding chapters, and it is likely that the longer separations are merely the result of a combinations of such factors. An alternative and potentially complementary view is that long separations can be exploited to manage the audience’s expectations, an idea that I evaluate in relation to my data.