Bryn Mawr Classical Review 95.11.08

Helma Dik, Word Order in Ancient Greek: A Pragmatic Account of Word Order Variation in Herodotus. Amsterdam Studies in Classical Philology 5. Amsterdam: J. C. Gieben, 1995. Pp. xii + 294. f. 143.10. ISBN 90-5063-457-5.

Reviewed by David Sansone, University of Illinois (

Arthur Sidgwick gave the following advice to the boys (!) to whom his Introduction to Greek Prose Composition is addressed: "The chief thing to remember about the order in Greek prose sentences is that it is the natural order." It was of course easy for Sidgwick to say this, as he had more or less successfully interiorized the "rules" for Greek word order as a result of his long years of reading and composing Greek (and correcting the compositions of his "boys"). Every language has its own preferred order for the placement of words and constituents, and those who have mastered a given language "naturally" adhere to that preferred order. But, when it comes to the practical realities, what exactly are the "rules" for word order in ancient Greek? A number of attempts have been made over the years; until now the most successful effort has been that of Sir Kenneth Dover, whose Greek Word Order of 1960, however, has not been as influential among classicists as it might have been, in part, perhaps, because of the presence in it of statements like the following (47): "Here FASI/N is strictly speaking Mq in character, and the word-group which I have analysed as N Cq C is therefore on the borderline of the category 'C group'; it admits of the analysis N Mqq C." It was in order to avoid having to read sentences like this that many classicists became classicists in the first place. Still, Dover's book is of fundamental importance, and it includes a comprehensive bibliography of earlier work. In his Preface Dover gracefully acknowledges his debt to his predecessors, saying of his slim volume: "... whatever good there may be in it has been reached by standing on the shoulders of others. I hope that someone will stand on mine as soon as possible, and that I can take the weight." Rather later than sooner, as it turns out, Sir Kenneth's wishes have been answered. Whether he can take the weight remains to be seen; he will certainly appreciate the irony of his supercessor's surname. Be that as it may, Word Order in Ancient Greek is a substantial and impressive contribution to the study of the Greek language.

The approach taken by Dr. Dik in this, her dissertation at the University of Amsterdam, is one that was not available to Dover in 1960. Her approach is that provided by Functional Grammar, which was developed, beginning in the 1960s, by Simon Dik, whose recent and untimely death represents a great loss to the community of linguists. (Helma Dik, I am told on good authority, is not related to Simon.) Most Hellenists will not be familiar with Functional Grammar, but that is no impediment to the appreciation of this lucid study, for Dr. Dik is a remarkably accommodating companion. In addition to writing in flawless and idiomatic English, she keeps technical terms to a minimum. Her definitions of such technical terms as are necessary for her presentation are models of clarity and concision, and the reader is further helped by a convenient "Index of Terms" that makes it easy to locate those definitions. The two most important terms in her discussion denote the pragmatic functions of Topic and Focus, which are defined and illustrated on pp. 24 ff.: "In the theory of Functional Grammar, we can identify the notion of Topic with the information that serves as a point of orientation, and Focus with the most salient piece of new information in a clause." The main purpose of the book is to make sense of the distribution of Topic and Focus within the clause, and the pattern that Dik proposes (p. 12) is as follows: P1--PØ--V--X. In this pattern, "P1" = the slot for Topic; "PØ" = the slot, immediately preceding the verb, for Focus; "V" = the "default position for the verb" in those cases where the verb has neither Topic nor Focus function; and "X" = the slot for everything else (i.e. those non-verbal elements that are not assigned a pragmatic function). This scheme is illustrated and defended in brilliant fashion from the text of Herodotus. Naturally, there is a danger in confining oneself to an examination of the language of a single author: The possibility exists that what Dik has discovered is not a pattern of Greek word order, but a pattern of Herodotean word order. But this can be tested. The testing will need to be done over an extended period of time and using a variety of texts. My own preliminary (and rather desultory) investigations reveal that, at the very least, this pattern holds for Xenophon and Lysias as well. In other words, there seems to be a good possibility that Dik has described for the first time the character of ancient Greek word order. This is no small feat, particularly when one considers the fact that grammarians have been attempting to do this very thing since the time of Dionysius of Halicarnassus.

I hasten to add that Dik has not merely taken over the theory of Functional Grammar and applied it mechanically to ancient Greek. In one important respect she has suggested (and successfully defended) a modification in the standard theory of Functional Grammar. In an extended discussion (64-70, 207-35) she demonstrates that, contrary to the orthodox view, predicates, as well as terms, can receive Topic assignment. Dik's demonstration is entirely convincing although, as she herself recognizes (208 n. 200), topicalization of the predicate is a characteristic feature of Herodotean style. (Dik's bibliography unfortunately does not include Mabel Lang's Herodotean Narrative and Discourse, the first chapter of which provides an interesting analogue on the narrative level for this verbal characteristic.) But one does not have to read very far in Xenophon's Anabasis before one stumbles upon a cluster of instances which show that this is not an exclusively Herodotean feature and which illustrate perfectly Dik's claim. At An. 1.2.14-16 we find LE/GETAI, E)KE/LEUSE, E)TA/XQHSAN and E)QEW/REI, all heading clauses and all serving as Topic for the clauses in which they appear. Particularly neat is the way in which E)TA/XQHSAN, following as it does TAXQH=NAI and SUNTA/CAI, exemplifies Dik's statement (209) that "perhaps the most obvious candidates for Topic function are repeated predicates, referring to an action mentioned already, but adding additional information." The remaining three Topics fall readily into Dik's category of "inferrable predicates," which is explained on pp. 215 ff., a passage that well displays Dik's standing as both a canny linguist and a sensitive philologist. The way in which Dik has gone about demonstrating the validity of the pattern P1--PØ--V--X is particularly well conceived. She has chosen to investigate all the instances in Herodotus of a select group of predicates: in Chapters 4 and 5 the verbs STRATEU/OMAI, A)/RXW and BASILEU/W; in Chapter 6 four verbs of speaking in the third person singular. (The choices seem to have been dictated in some measure by a spirit of playful engagement with Dover; see Greek Word Order 25-26, 49-50, 53.) This procedure, which is justified in a few lucid paragraphs (15-17), serves to minimize the number of variables and allows for the discovery, not exactly of minimal pairs,1 but of "slightly-more-than-minimal sets of instances." So, for example, the difference between OI( ME/N NUN A)/LLOI TRIH/REAS PAREXO/MENOI E)STRATEU/ONTO and DO/NTES DE\ KAI\ *KU/PRIOI SFE/AS AU)TOU\S *PE/RSH|SI E)STRATEU/ONTO can be readily appreciated and is convincingly explained (80) by the fact that in the former OI( A)/LLOI is Topic and TRIH/REAS is Focus, whereas in the latter DO/NTES is Topic and KU/PRIOI is Focus. Another fine example of the advantages of this procedure is the discussion (95-100) of preposed vs. postposed first arguments of the verb BASILEU/W , i.e. cases where, in essence, the grammatical subject precedes the verb vs. cases where it follows. The instances show -- and this seems to be fully borne out by the non-Herodotean material that I have reviewed -- that preposed first arguments are pragmatically marked (i.e. serve either as Topic or Focus), while postposed first arguments present "non-vital" information. Indeed, one of the most acute observations that Dik makes -- again, this is borne out by independent observation -- is that there is a very strong tendency in a Greek sentence for "predictable" material to follow the predicate, a tendency that may receive some confirmation (5 n. 8, 35 n. 64) from the tapering off of intonation through the course of the clause. A further benefit of Dik's approach is that it enables her to account for phenomena that have already been observed by traditional grammarians -- or, as she engagingly calls them (53), "sentence grammarians," as opposed to the "discourse grammarians" of which she is a radiant example. It has previously been noted, for instance, that the order subject-verb-object is a common, if not the most common, order in ancient Greek. In the course of a brilliant paragraph (93-94; cf. also 102, 256 n. 252) Dik explains, in effect, why this is the case in pragmatic terms.2 Another good example of the explanatory force of Dik's method is to be found on p. 216: Whereas traditional grammar, when confronted with Herodotus' TOU= ME\N A(MARTA/NEI, TUGXA/NEI DE\ TOU= *KROI/SOU PAIDO/S, merely affixes a label ("chiasmus"), Dik manages to give us an account of what in fact is going on.

An index of the value of Dik's book is the number of incidental bits of information one picks up along the way. For example (79), "a check on all collocations of A)NH/R and A)GAQO/S in Herodotus shows that A)GAQO/S never precedes A)NH/R." Indeed, this seems to be the case in prose in general (except, of course, when A)GAQO/S is in attributive position: O( A)GAQO\S A)NH/R), so that the order A)NH\R KA)GAQO/S seems to be as fixed as the order KALO\S KA)GAQO/S. We learn, further, that "it is extremely rare for clauses to have two full NPs [= noun phrases] at one side of the verb, without accompanying participial phrases" (206 n. 196). This observation should be considered in conjunction with the excellent discussion (23-24) of "structure-building," particularly the requirement that information be presented in "manageable chunks." Again, I have not seen it pointed out previously that prepositional phrases with (anaphoric) AU)TO/N etc. are postpositive.3 And the observation that "using a less specific term for an established participant or State of Affairs ... is a well-known fact of D-Top [= Discourse-Topic] expression" (213) needs to be taken into account by those who discuss the phenomenon of "compound-simplex iteration."4 Finally, in the course of her investigation of different verbs for speaking, an important distinction in use between the aorist and imperfect is drawn (167): "Contrary to the 'reacting' EI)=PE, E)/LEGE usually opens a discussion ..." I have checked this against a text that I thought I knew pretty well, Plutarch's Regum et imperatorum apophthegmata, which contains what must be the highest concentration of verbs for speaking in all of Greek (EI)=PE: 183; E)/LEGE: 44). While this distinction is not absolute, it turns out that it holds in the vast majority of instances, and Dik's discussion (esp. 136, 164-70) provides a valuable supplement to Rijksbaron's "The Discourse Function of the Imperfect," In the Footsteps of Raphael Kühner (Amsterdam 1988) 237-54 and to H. Fournier's Les verbes "dire" en grec ancien (Paris 1946).

There is little indeed to criticize in this illuminating book, particularly since Dr. Dik, with refreshing candor, herself frequently acknowledges those problems that her treatment is unable to resolve. Only very rarely -- and usually concerning relatively minor matters -- does her judgment deserve to be questioned. For example, Dik defends the anomalous position of Topic in STRATEU/ONTAI W)=N E)P' AU)TOU\S by pointing out that E)P' AU)TOU\S W)=N STRATEU/ONTAI would be "ungrammatical" (62). But what was to prevent Herodotus from writing E)PI\ TOU/TOUS W)=N STRATEU/ONTAI? And Dik's inclination to follow Schaefer (sic) in regarding as an interpolation an instance that she finds awkward (174) eliminates the anomaly from the text of Herodotus, but not from the body of ancient Greek produced by native speakers. But Dik has undertaken an ambitious project, and she will be the first to admit that a great deal of work still needs to be done. Indeed, her "Conclusion" to Chapter 8 contains a very sensible and modest appraisal of just how far her approach has taken us and what additional areas must still be investigated. A generation elapsed between the publication of Dover's book and that of Dik's. It is to be hoped that we will not have to wait yet another generation before someone tests the strength of Dik's shoulders.


  • [1] Note, however, the discussion (156-57) of EI)=PE PRO\S TOU=TON and PRO\S DH\ TOU=TON EI)=PE. Dik's explanation of the variation in word order here cannot be improved upon. I must admit, however, that no reasonable explanation seems forthcoming for the variation (161) between E)/LEGE TA/DE and TA/DE E)/LEGE.
  • [2] In so doing, Dik confirms the statement with which Dover concludes his chapter on syntactical determinants (Greek Word Order 31), that statistical data regarding the position of subject, verb and object "suggest with increasing force that all patterns of order which are describable in syntactical terms are secondary phenomena."
  • [3] See 32 n. 58, 62, 138, 153-54, 252. I am inclined, however, to disagree with Dik (186 n. 193) and to view Hdt. 9.119.2 as a counter-example. The same is true of 7.10Q.2 PRO\S DE\ AU)TOI=SI KAI\ E)GW/ (not mentioned by Dik), although it must be admitted that such prepositional phrases in initial position are quite rare.
  • [4] See R. Renehan, Studies in Greek Texts, Hypomnemata 43 (Göttingen 1976) 11-22; for more recent bibliography, see J. Diggle, Euripidea (Oxford 1994) 84 n. 64.