This book examines the use of Greek words or phrases in Latin letters, or in linguistic terms code-switching; it is not a study of epistolary language more generally, though a number of the conclusions have broader application. It includes chapters focusing on Cicero, Fronto, and Suetonius (primarily Augustus’ letters), preceded and followed by more general discussions. Mullen is responsible for the two introductory chapters and the one on Fronto, and Elder for the concluding chapter and those on Cicero and Suetonius — but readers would not be able to spot these divisions if the authors had not told us, for the work is seamlessly integrated in both style and content.
Cicero’s famous code-switching has already been extensively studied, including by J. N. Adams,[1] and readers will no doubt be sceptical about whether anything remains to be said. Nevertheless Elder and Mullen have indeed managed to provide contributions there as well as for later authors, for example making interesting additions to Adams’ observations about the reasons why Cicero stopped using code-switches at certain key periods in his life (pp. 136-43); the chapter on Cicero also offers a clear and sensible overview of earlier work and is likely to replace much of it. Overall, interesting findings include that Romans explicitly debated both the status of particular Greek words in Latin and who had the right to determine that status, i.e. to ‘give citizenship’ to a foreign word; that owing to the Romans’ complicated relationship with Greek, code-switches could function both as ‘marked’ (i.e. striking, unusual) language and as ‘unmarked’ language; that attitudes to the use of Greek both changed over time and varied between individuals; and that the language of Roman epistolography was ‘built out of a range of different languages, oral and not, popular and elite, Greek and Latin, and reflects the broader continua of Roman communication and society.’ (p. 285)
The conclusions are based on a corpus of 1,403 code-switches (981 from Cicero’s letters, 141 from Fronto’s letters, 142 from Pliny’s letters, and 139 from Suetonius, in the last case not only from letters) which have been entered into an open-access database. The database not only provides evidence to back up the claims made in the book, but also constitutes a valuable resource for future work on Roman code-switching. It gives the references for each code-switch (including, for Cicero, both the traditional letter numbers and Shackleton Bailey’s numbers), its date (if known), its source (i.e. whether it comes from a letter or not; as only the Suetonius examples can come from sources other than letters, the prominence of this information in the entries from other corpora is arguably not a good use of prime space), its author (all three epistolary corpora include letters sent to as well as by the person after whom the corpus is named), its addressee, its exact form, its syntactic status (i.e. intra-sentential when a short piece of Greek is incorporated into a Latin sentence, or inter-sentential when the Greek and Latin are syntactically separate), its function (e.g. evoking the Greek cultural sphere (‘GCS’), quotation of a literary or non-literary source, citing a word under discussion, insults, exclamations, wordplay, etc. On pp. 25-9 the full list is given with an example of each, while the appendix on pp. 291-307 provides more information on how often each function occurs in various contexts, and sometimes other information, such as the Latin context of the code-switch (cf. pp. 20-2). The database, which can be both browsed and searched, is very user-friendly, and a spot check of entries suggests that it is also accurate (the ‘function’ category is highly subjective, of course, but the authors were aware of that and took care to ensure internal consistency in the use of different categories). It contains much more information than is in the book, particularly in the case of Pliny, whose letters are not treated in the book in any detail, and it could provide the raw material for several articles at least.
Code-switching has been a hot topic in modern linguistics for several decades, so that the sheer quantity of linguistic work needing to be absorbed (as well as its varying quality) poses a challenge for Classicists who want to investigate ancient code-switching from a theoretically informed perspective. Elder and Mullen make a judicious if minimalist selection of the linguistic literature and discuss it intelligently;[2] their discussion may well become the first port of call for Classicists interested in the theory of code-switching, as it is clearer and more sensible than most other treatments. Linguists, however, will probably find it unfortunate that Elder and Mullen’s selection of linguistic theories is explicitly based in part on the choices of earlier Classicists (p. 63) and thus contributes to a growing pattern whereby Classicists’ use of linguistic theory becomes more and more of a separate theoretical world, divorced from the mainstream of modern linguistics. There is a sharp divide between Elder and Mullen’s treatment of linguistics and their treatment of history, where they clearly make their own choices of the most relevant scholarship; one suspects that they would not dream of suggesting that those choices ought to be limited to the theories of ancient history that had been used by previous scholars working on these letters.
A code-switch is distinct from a loanword or borrowing, that is a word of foreign origin that has become accepted in another language. In written English code-switches are normally marked with italics or a change of alphabet, while loanwords are not so marked. But there is inevitably a grey area of words that are partially or sometimes accepted into the borrowing language, ones that in English might be italicized by one writer and not by another. For a study with a statistical dimension, the question of how to identify code-switches in Latin is crucial, but at the same time not easily soluble. Many words of Greek origin were borrowed into Latin, and while the use of Greek endings and Greek script can act as indications of a word’s foreignness, words that appear with Latin endings and in Latin script may nevertheless be explicitly commented on by ancient authors as being foreign. To make matters worse, we cannot have any confidence that the script choices in our manuscripts go back to those of the letters’ original composers. Elder and Mullen discuss this situation candidly and offer reasonable justification for their choice, in the case of Cicero’s letters, to consider as code-switches only those words that Shackleton Bailey prints in Greek script (pp. 120-5). But for Fronto and Suetonius they consider as code-switches numerous words that editors print in Latin script (pp. 188, 241). A justification for the different policy is lacking, as is an explanation of the criteria used to identify code-switches in Latin script. Here more engagement with the work of modern linguists would really have been helpful, as would consideration of usage in Latin documentary texts (questions like ‘did ancient authors themselves engage in script-switching?’ [p. 120] cry out for a consideration of the evidence of papyri, ostraca and tablets).
The clarity of the exposition is impressive, particularly given the complexity of the issues involved: theories are carefully explained when introduced, and technical terminology is scrupulously defined when first used. Most readers will, however, need to read the work through from the beginning to understand it. Those who tackle a later chapter in isolation will especially appreciate the excellent index, which points to the explanations of key terminology, to mentions of the most important modern scholars, and to famous code-switches, etc. The frequent Latin quotations are almost always translated, but Latinless readers will still have difficulty understanding exactly how the Greek words fit into their Latin frames. On pp. 266-7 bold font is used to mark the original Greek in the English translation; it is a pity that this useful device is not employed in the rest of the book.
Despite its many merits, the work has some annoying features. Sweeping generalizations are sometimes made on debatable points,[3] either without any supporting evidence or with reference to only one or two works of modern scholarship whose authority readers may not be inclined to accept unquestioningly; often a proper discussion is offered later, but not always, and even when one is offered readers may be frustrated by being asked to take things on trust earlier. When arguments are spelled out in more detail, they are occasionally unfair, for example on p. 89 where the finding that over 41% of Fronto’s code-switches are literary forms the basis for a statement that ‘Fleury is wrong about code-switching in Fronto regularly being used outside the literary sphere’ (surely 59%, or even considerably less, could count as ‘regularly’?).[4] There is a pervasive tendency to frame statements as being primarily about the authors’ database and only obliquely about the ancient texts themselves;[5] while it is excellent that this database is available, it is a tool for the study of ancient letters, not an object of study in itself. Page numbers are sometimes omitted from references.[6] There are some typos, including in the Latin and Greek[7] — but not many, for in general the standard of production is high.
Overall, despite a few irritating features, this work makes excellent contributions to the study of ancient bilingualism and will be useful to scholars interested in Cicero, Fronto, Augustus, or Suetonius from any perspective, as well as to those with more specifically linguistic interests.
Notes
[1] J. N. Adams, Bilingualism and the Latin Language (Cambridge 2003) pp. 308-47.
[2] It is unfortunate, however, that on p. 188 the concept of ‘nonce borrowings’ is suddenly thrown into the theoretical mix; that concept belongs to a minority view and is not completely compatible with some of the other theoretical material used. It is also unfortunate that this point is accompanied by a vague cross-reference to Chapter 1, where there does not seem to be any discussion of nonce borrowings.
[3] E.g. ‘Any “naturalism” we see is still a deliberate choice: the code-switching is carefully policed, only occurring between certain correspondents under certain conditions.’ (p. 93)
[4] Likewise on p. 105 ‘we might even claim that a defining characteristic of high-status Romans was full competence in both Latin and Greek linguistic and cultural spheres’: this suggestion is not supported by the evidence offered. On p. 250 a series of appropriately cautious statements about how Suetonius ‘may have’ presented an unrepresentative sample of Augustus’ epistolary style seems then to be taken as proof that he has done so, when the argument continues with ‘Further distortion comes from…’ On p. 130 an argument about intra-sentential switching not being as closely connected to intimacy as normally thought seems to confuse the amount of intra-sentential switching (the factor most likely to be linked to intimacy) with the percentage of switches that are intra-sentential (a figure heavily influenced by how many literary quotations an author uses, since quotations tend to be inter-sentential).
[5] Information on how often particular features occur in the database (often provided) is of course very helpful, but references to the database are distractions in statements like ‘The five examples of imagined quotations in Greek found in Fronto’s correspondence are all included in the database with more than one function’ (p. 202). What matters is that these examples have multiple functions in their ancient contexts, and that is why the authors so categorized them when constructing the database.
[6] E.g. on p. 58 n. 36, in reference to a book of more than 800 pages.
[7] E.g. ιυντάξομαι for συντάξομαι (p. 164), officum for officium (p. 116).