Bryn Mawr Classical Review

[The table of contents is given at the end of the review.]

Greek courses teach us early on that adjectives and their nouns can go in either order (ἀνὴρ ἀγαθός or ἀγαθὸς ἀνήρ 'a good man'), and that if there is a definite article involved, there are various possible positions for attributive adjectives: between the article and the noun (ὁ ἀγαθὸς ἀνήρ 'the good man') or after the noun, with a repetition of the article (ὁ ἀνὴρ ὁ ἀγαθός 'the good man'). More rarely, a noun without a preceding article is again followed by article plus adjective (ἀνὴρ ὁ ἀγαθός 'the good man'). Predicative adjectives go either before or after the article+noun group (ἀγαθὸς ὁ ἀνήρ or ὁ ἀνὴρ ἀγαθός 'the man is good'). What is usually less clear is what difference, if any, it makes to say ἀγαθὸς ἀνήρ rather than ἀνὴρ ἀγαθός, or ὁ ἀνὴρ ὁ ἀγαθός rather than ὁ ἀγαθὸς ἀνήρ. And we have not even mentioned nouns modified by more than one adjective, or nouns modified by dependent genitives, prepositional phrases, relative clauses, and so on.

This book is about the order of elements and the use of articles within the noun phrase (i.e. noun together with its modifiers) in Herodotus. It has two parts. The first treats the ordering of the noun relative to its modifier(s), and the ordering of modifiers relative to one another. The second treats the circumstances under which noun phrases have an article at all, and the various positions where articles can appear in noun phrases. Part 1 comes first because the relative ordering of nouns and their modifiers is found to be determined by factors largely independent of the use of articles in noun phrases (although this conclusion has to be taken on trust to begin with: see p. 33).

The main conclusions of part 1 may be summarised as follows. For noun phrases with single modifiers, the modifier precedes the noun if it is presented as, roughly, the most important part of the noun phrase in context; Bakker's preferred term for contextual importance is 'salience'. The noun precedes the modifier if neither is singled out as salient, or if the noun is more salient than the modifier. These tendencies are sometimes overridden by various other factors, but remain the most common determinants of ordering within the noun phrase. The main principles here are those that Dik (1997) proposes, also on the basis of Herodotus, for nouns modified by adjectives (cf. Dik 2007: 84-122), but Bakker extends the argument to all kinds of modifiers and bases it on a much larger collection of evidence. For noun phrases with multiple modifiers, Bakker extends the principle that the most salient elements come first so that the noun and its various modifiers are all ordered from most to least salient. Thus at Herodotus 3.130. 4, Darius presents Democedes πεδέων χρυσέων δύο ζεύγεσι 'with two pairs of golden fetters'. The dependent genitive πεδέων χρυσέων is given most salience as the essential description of the gift, and δύο is given more salience than ζεύγεσι to prepare for a joke depending on the concept of 'two' (pp. 109-10).

In part 2, Bakker rejects recent interpretations of the Greek article as a device for marking the topic of a sentence. Instead, she argues that the Greek article functions like definite articles in many other languages. Her preferred account of definite articles in general, and in Greek, is a variant of the view that definite articles mark entities out as identifiable: 'a definite article is appropriate if the speaker presents the referent in question as unequivocally relatable to an available cognitive structure that is relevant in the given discourse' (p. 162).

The real heart of part 2 is the chapter on the placement of articles in noun phrases. Here Bakker distinguishes between types of modifiers for which the so-called 'predicative position' (ὁ ἀνὴρ ἀγαθός or ἀγαθὸς ὁ ἀνήρ) really indicates predicative function, and other types of modifiers (e.g. the dependent genitive in τῆς τάφρου τὰ χείλεα 'the border of the moat', Herodotus 1. 179. 2). Adjectives, numerals, and sometimes participles fall into the first category, other modifiers into the second. For modifiers of this second kind, Bakker argues that the modifier is immediately preceded by an article (one agreeing with the noun in number, gender and case) if this modifier helps to clarify which entity the noun phrase refers to, and not otherwise (see p. 226). Thus at Herodotus 7. 194. 1 (see pp. 238-9), the crews of fifteen of Xerxes' ships spot τὰς ἐπ' Ἀρτεμισίῳ τῶν Ἑλλήνων νέας 'the Greek ships at Artemisium'. Here the first modifier ἐπ' Ἀρτεμισίῳ clarifies which ships are intended; the second τῶν Ἑλλήνων does not help to distinguish between different groups of ships: it only adds further information about the ships at Artemisium. This does not mean that τῶν Ἑλλήνων is unimportant (in which case it should follow the noun); Xerxes' men turn out to be fatally unaware that these ships are Greek. At Herodotus 8. 81 Aristides is reported as saying that the Greek camp was entirely surrounded ὑπὸ τῶν νεῶν τῶν Ξέρξεω 'by Xerxes' ships'. Here the modifier Ξέρξεω is preceded by an article because it helps to clarify which ships are intended. But the modifier follows the noun because Aristides' (and Herodotus') audience hardly needs to be told that the ships in question are Xerxes': the modifier merely confirms that this is the intended referent (p. 234).

A noun followed by an articular modifier tends to receive an article itself (τῶν νεῶν τῶν Ξέρξεω), but does not always. Bakker argues that the noun receives its own article if the information provided by the modifier is helpful but not essential for the interpretation of the noun phrase (see p. 277). When the noun is left without its own article, the information provided by the modifier is essential. Herodotus (6. 138. 2) says that the Lemnians' Athenian concubines taught their children γλῶσσάν τε τὴν Ἀττικὴν καὶ τρόπους τοὺς Ἀθηναίων 'the Attic language and the manners of the Athenians'. Here the lack of article with γλῶσσαν and with τρόπους tells us not to jump to conclusions about what language and what manners the children were taught: we have to wait for the modifiers. If one wonders why these crucial modifiers follow their nouns, the answer is that the nouns themselves are contrastive, and therefore even more salient (p. 280, n. 91).

When a noun is preceded by article+modifier (τὰς ἐπ' Ἀρτεμισίῳ νέας), Bakker takes the article to belong to the modifier rather than to the noun (just as she does when article+modifier follows the noun). A noun preceded by article+modifier never gets its own immediately preceding article (τὰς ἐπ' Ἀρτεμισίῳ τὰς νέας is not a pattern that occurs). Bakker suggests that 'as articular modifiers only occur in definite NPs there is no need to mark the noun for definiteness as well' (p 284)--a slightly odd suggestion given Bakker's view that, when an articular modifier follows, articulation of the noun does more than mark the noun phrase as definite. A suggestion more consistent with Bakker's overall thesis might be that, since preposed modifiers are always salient, there is no such thing as a noun phrase with a preposed but non-crucial 'articular modifier'.

Finally, if a noun phrase is definite but none of its modifiers (if any) receives an article, the noun gets an article as the default way of indicating definiteness.¹

Part 1 is based on a study of all Herodotus' attributive uses of all modifiers (although noun phrases involving certain kinds of noun are excluded: see p. 4). Part 2 is based on all the noun phrases with common nouns in books 2 and 7. These substantial collections of evidence are very welcome, yet it is impossible for Bakker to discuss more than a small part of this evidence. Resisting the temptation merely to select convenient examples, she illustrates each point first on the basis of some very nice examples; she moves on to more difficult but still tractable examples, and then goes on to say how many intractable exceptions she found and to lay these out. What we do not get, however, is a good sense of what a really representative sample of her evidence would look like: all the examples in a particular chunk of Herodotus, or all those involving particular nouns or modifiers. If we are really to evaluate Bakker's hypotheses, the onus is on us to read Herodotus (and other authors) ourselves and see how well we think Bakker's principles apply.

The book is refreshingly candid throughout: when Bakker cannot explain something she says so. In this spirit, she notes not only the unavoidable subjectivity in the contextual analysis her hypotheses require, but also the problem of circularity 'which may arise if we interpret our data according to the principle we want to establish and subsequently use these interpretations as evidence for the principle' (p. 32). As she says, the reader has to decide: but the reader is given many attractive hypotheses to evaluate.

This review cannot do justice to the numerous subsidiary points made over the course of the book. A highlight, to my mind, is the point that predicate nouns have an article if they are 'identifying' rather than 'classifying' predicates (pp. 190-7; cf. Procksch 1881). I am less convinced that noun phrases such as οὗτος ἀνήρ (with demonstrative but without article) are indefinite (pp. 182-9), a conclusion that leaves Bakker struggling later to get around a difficult contradiction (p. 263). But the point is important if correct, and deserves to attract further debate.

One might quibble that important parts of the argument are sometimes introduced in footnotes,² and other scholars' views are not always treated as carefully as they might be.³ But an impressive amount of analysis of the text of Herodotus has gone into this book, and the results deserve attention.

Bergson, L. 1960. Zur Stellung des Adjektivs in der älteren griechischen Prosa. Stockholm: Almqvist and Wiksell.

Devine, A. M. and L. D. Stephens. 2000. Discontinuous Syntax: Hyperbaton in Greek. New York: Oxford University Press.

Dik, H. 1997. 'Interpreting Adjective Position in Herodotus', in E. J. Bakker (ed.), Grammar as Interpretation: Greek Literature in its Linguistic Contexts (Leiden: Brill), 55-76.

Dik, H. 2007. Word Order in Greek Tragic Dialogue. Oxford: Oxford University Press.

Procksch, A. 1881. 'Ueber den Gebrauch des Artikels, insbesondre beim Prädicat', Philologus 40: 1-47.

Slings, S. R. 1997. 'Figures of Speech and Their Lookalikes: Two Further Exercises in the Pragmatics of the Greek Sentence', in E. J. Bakker (ed.), Grammar as Interpretation: Greek Literature in its Linguistic Contexts (Leiden: Brill), 169-214.

Table of Contents

1.   The chapter on the placement of articles in noun phrases is based on a collection of definite noun phrases, i.e. noun phrases in which an article appears. It is inevitable, therefore, that in these noun phrases 'the noun is always preceded by an article' if there is no modifier preceded by an article (p. 284). Nevertheless, the point that nouns may receive an article purely to mark the noun phrase as definite is non-trivial (since Bakker could have found that no articular nouns in her corpus required this explanation).
2.   Thus p. 53, n. 27 introduces the idea that noun-modifier order is the default order for single-modifier noun phrases, occurring when neither noun nor modifier is singled out as salient.
3.   Bakker (p. 20) notes that Devine and Stephens (2000: 21) calculate that 'although both determining and qualifying adjectives may be pre- and postnominal, there is a highly significant correlation between determining adjectives and prenominal position'. She then states (p. 21, n. 24) that Dik (2007: 85) "must be confusing determining adjectives with qualifying ones when she says that: 'I take it that determining adjectives are simply less likely to constitute the most salient part of a noun phrase than qualifying or quantifying ones (my italics)'". In fact, Dik means what she says here. Although Devine and Stephens (2000: 21) do indeed calculate, for Herodotus, that determining adjectives are found before their nouns much more often than qualifying adjectives, they go on to argue that the opposite correlation is found in Thucydides. Bergson (1960) had previously argued for classical prose in general that determining adjectives, much more than qualifying adjectives, usually follow their nouns. It is this more general claim that Dik (2007: 85) responds to.

At p. 58, n. 30, Slings' (1997: 184-92) view of chiasmus is discussed in such a way that it would appear that at Odyssey 13. 8-9 (ὅσσοι ἐνὶ μεγάροισι γερούσιον αἴθοπα οἶνον | αἰεὶ πίνετ' ἐμοῖσιν, ἀκουάζεσθε δ' ἀοιδοῦ), Slings considered οἶνον to have Focus function in the first clause and ἀκουάζεσθε to have Focus function in the second. While Slings (1997: 189) did indeed consider οἶνον to be part of the Focussed constituent in the first clause, he considered ἀκουάζεσθε to have Topic function in the second.