It takes only a passing familiarity with Classical Greek as well as many other old Indo-European languages to be aware of the rule that certain small words, that is, ‘particles’ such as μέν and δέ, are wont to come ‘second’ in their clause. This phenomenon has become known as ‘Wackernagel's law’. However, the specification of exactly what it means to be ‘second’ is, on closer inspection, harder than one might think to establish (see e.g. J. Clackson, Indo-European Linguistics: An Introduction [2007], pp. 168–9). This is part of the much broader problem of accurately describing surface word order in Greek, which from antiquity to the present has been notorious for its ‘freedom’. Coming in a line of recent monographs focusing on work on surface word order in Ancient Greek (including H. Dik, Word Order in Ancient Greek [1995] and Word Order in Greek Tragic Dialogue [2007], and S. Bakker, The Noun Phrase in Ancient Greek [2009]) as well as Greek historical syntax more generally (e.g. P. Probert, Early Greek Relative Clauses [2015]), G.’s study makes a critical contribution to our growing understanding of the conundrum of Greek word order and surface syntax. The monograph provides key insights not only into the core question of the behaviour of clitics, but as a result of investigating a phenomenon that lies at the interface of phonology and syntax, it also has significant implications for general linguistics. For the Classicist its primary value will be the implications of G.’s research for the interpretation of texts.

G. lays the theoretical foundations for his work in Part 1, which comprises three chapters on Greek syntax and surface word order, the prosody of Greek clitics and the syntax of clitics respectively. Here he establishes that Greek is a discourse-configurational language, which is to say that discourse and information structure are, to a much greater degree than in many modern western European languages, critical determinants of the surface word order. While Wackernagel's law targets particles operating at each of sentence, clause and phrase levels, G.’s focus is on clause-domain particles, such as ἄν and pronominal clitics serving as verbal arguments, for example μιν (p. 9). He finds that clause-domain clitics select a prosodic word, as distinct from a syntactic word, to serve as their host (p. 84). Crucially, the host of a clausal clitic need not be a syntactic constituent, which might be expected if syntactic constituency were the critical factor, as it is in some languages such as Czech, for example ἐπὶ κόσωι ἂν χρήματι (Hdt. 3.38.3), where ἂν occurs after κόσωι rather than ἐπὶ or χρήματι (pp. 68–71). However, this is not to say that syntactic factors are irrelevant: G. finds that the scope of the clitic in question critically affects its positioning, so that ἄν as a modal particle behaves differently from ἄν as a domain-widener (p. 118).

Having laid the foundations in Part 1, G. moves on in Part 2 to address cases where the preposing of phrases owing to topicalisation or focus leads to clausal clitics occurring in a position other than second in their clause. Topicalisation is used to manage transitions between discourse referents (p. 172). By contrast, focus preposing is used to ‘assert a value for a proposition that already has a value’ in the discourse (p. 12). Topicalised and focus preposing phrases occupy different positions, with the former positioned to the left of the latter. What these constructions have in common is that they place material to the left of the host of a clausal clitic, as in the following example of topicalisation from Hdt. 3.92.1 (p. 129, text & trans. G.), where the phrase given in square brackets is topicalised, and the clausal clitic οἱ is hosted by the next word, χίλια (square brackets denote the topicalised phrase):

These observations open up exciting possibilities for those interested in interpreting texts. Specifically, the preposing of the topicalised phrase in this case, as evidenced by the position of the clausal clitic οἱ, allows for the overt identification of a topicalised phrase, and thus the position of the sentence within the wider discourse structure. In this way G. has given readers of Greek a tool for identifying the topic and focus of a clause, which, in the absence of a clausal clitic, would not necessarily be obvious.

In the final part G. moves on to treat non-finite constructions, involving participles and infinitives. In both cases G. uncovers two fundamental types determining the positioning of clausal clitics, which he terms S- and VP-constructions. In VP-constructions the participial/infinitival phrase is closely tied to its matrix finite clause, whether, for example in the case of the former, by modifying an element of that clause, or, in the latter, by constituting the embedded infinitive in a control predicate. These constructions do not provide an independent domain for clausal clitics, and such clitics are hosted by the first prosodic word of the matrix and dependent construction taken as a whole (that is, they are treated as a single S/CP constituent, in G.’s terms). By contrast, S-constructions may be said to be free-standing clauses with respect to their matrix, as in, for example, a genitive absolute construction or an infinitival clause introduced by a verbum dicendi. These cases constitute an independent domain for clausal clitics, and the latter will be hosted by the first prosodic word of the participial/infinitival clause. Once again, G.’s findings constitute a diagnostic tool for understanding the precise relationship between elements of the sentence in ancient Greek, and as such provide invaluable interpretative clues for the reader.

One feature of which the reader should be aware is that the transliteration of the Greek is not into the Latin alphabet, but into symbols of the International Phonetic Alphabet (IPA). This has no doubt been done for the excellent reason of giving the reader an understanding of the phonological values of the characters at the time of Herodotus, from whom the vast majority of the examples are taken. Nevertheless, it gives the impression that what is given is somehow a representation of the original sound. However, the mapping is in fact to the Greek letters, that is, graphemes, not phonemes, the full scheme of which is given on p. xvi. Accordingly, /a/ and /aː/ are not distinguished, even though they could have been if the representation were phonemic. In fact, because the Greek script is largely phonemic, the difference is of little consequence in most cases. Where it is more of an issue is in the inscription IG I³ 755 from Athens (p. 67), whose pre-Euclidean script does not distinguish vowel length. Accordingly, <ε> and <ο> are represented as short even where they represent the phonologically long vowels that would at a later stage be written as <η> and <ω>. Thus ‘δεκατεν’ is given as ‘dεkatεn’, rather than ‘dεkatεːn’, which at first sight is a little confusing.

G.’s monograph represents a considerable achievement and is well worth reading for anyone interested in the interpretation of Greek texts, in the linguistics of ancient Greek or in Indo-European/general linguistics. Furthermore, G. has gone to considerable lengths to make the text accessible to both Classicists and general linguists, by providing examples in the Greek alphabet as well as in transliteration. Classicists may well not be familiar with the frameworks used, and some knowledge is assumed. Key terminology is, however, explained, and careful examination of the (very numerous) examples makes clear what is meant. The result is a very persuasive account of the placement of clausal clitics in Herodotus and, by extension, in ancient Greek as a whole.