Bryn Mawr Classical Review 2006.09.33
Andrew M. Devine, Laurence D. Stephens, Latin Word Order. Structured Meaning and Information. Oxford: Oxford University Press, 2006. Pp. xii, 639. ISBN 0-19-518168-9. $65.00.
Reviewed by M. Esperanza Torrego, Universidad Autónoma de Madrid (firstname.lastname@example.org)
Word count: 2717 words
This book is an ambitious study about one of the most difficult subjects in Latin grammar, Word Order (WO). WO is mainly conceived as a manifestation of the pragmatic meaning of a text. This pragmatic meaning is structured; it does not arise, thus, from the mere concatenation of discrete pragmatic units. This is the idea behind the subtitle of this work: Structured Meaning and Information. Additionally, WO is also triggered by other factors--semantic, syntactic, and sometimes prosodic. As a consequence, this work contains many interesting points that concern not only Pragmatics, but also Semantics and Syntax. This study covers WO of simple sentences and also of nominal phrases. Hyperbaton is included as an individual subject.
The investigation combines philological discussion of particular data with theoretical linguistic analysis within a Generative Grammar model (GG), mainly "a sanitized version of pre-Minimalist syntax with the addition of discourse functional projection", (p. 7-8). As a result, the book will be of interest to two kinds of readers: linguists working in GG, who have here for the first time, as far as I know, a general view of Latin WO within this framework, and Latinists and philologists in general, who may find a nicely detailed description of a great deal of data. Latinists working within the GG model are rare. Because the authors are aware of this, they present their work in a way that can be profitable for both groups of people: in each chapter they deliberately place the major part of the technical discussion in specific "Structural Analysis" sections. This way, readers can skip these sections and "still use the rest of the book as a largely descriptive account of Latin WO in the simple sentence" (p. 30).
The attempt to make the book accessible to a wide readership is welcome. However, no description is theoretically neutral and this is not either. Both the selection of data and the analysis are determined by the authors' theoretical approach. Overall, the book is full of interesting information about Latin grammar.
The study does not pretend to offer an exhaustive analysis of a corpus but uses Latin material in order to illustrate theoretically significant issues: "Our analysis proceeds mainly on the basis of probes or tests, using a few representative words in a restricted corpus of texts to build the elements of a general theory of Latin WO" ( p. 8). The only indication the authors give about the composition of the corpus is that it belongs to classical prose.
The structure of the book consists of an Introduction and six chapters (preceded by a preface, contents and abbreviations list), plus a bibliography, an index nominum and an index rerum. At the end of each chapter and before the endnotes, there is an un-annotated bibliography. It is not clear which criteria the authors have used for the selection of this bibliography.
In the Introduction (pp. 3-45) the authors present an overview of their starting points and methodological approach. Their main points are the following:
1) Latin WO is not random but reflects a particular speaker's aims, and it is necessary to capture a whole dimension of meaning.
2) Pragmatic meaning is compositional, not atomistic.
3) Syntax is the interface between semantics and pragmatics.
In such a model, the way in which syntax and pragmatics are connected is crucial. The basic points in this respect can be found on pages 25 and 26. They can be summarized as follows: syntactic structure is a hierarchical configuration in which a phrase of every category is a combination of a head (which represents the component of a category, Noun, Verb, Preposition) and its projection (argument, adjunct etc). In addition, the underlying structure of the sentence has three levels: the lower one represents the eventive structure; immediately above there is another functional projection that houses the subject and other potential displacement sites; finally, the third and highest level of structure is concerned with Focus and Topic.
The authors use this hierarchical structure to justify differences in Latin WO; for example, a complement of a head can appear in the basic complement position or moved to the higher structural level. Displacement operations in the second level are considered instances of scrambling. Scrambling is "a type of leftward movement in SOV languages. . . that has the potential to disturb the neutral order of arguments" (p. 136, n.1). WO in a sentence is determined primarily by discourse properties like topic and focus. Latin is said to be a discourse configurational language. A more detailed presentation is made in chapter 1, where the analysis is compared to other possible models (pp. 79-98 and 108-135). Within this framework, the surface of a pragmatically structured sentence is the result of applying to the neutral basic WO changes by scrambling and movements to the pragmatic functional positions. The aim of the book is to find out the "path covered" by phrases to reach the position they have on the surface and the factors that cause or favor such movements.
The departure point for the study of WO in sentences is defined in chapter 1 (pp. 36-144). It is devoted to sentences where the Subject is the topic and the material within the verbal phrase is new information. According to the authors, these sentences, called in the book "broad scope focus sentences," display a neutral or default WO. It is the following (p. 79):
Subject (S) -- Direct Object (DO) -- Indirect Object (IO) / Oblique argument -- Adjunct -- Goal or Source argument -- Nonreferential DO -- Verb.
Nonreferential DOs are accusatives that form with the verb a precompiled phrase, like bellum inferre, impetum facere. These kinds of DOs have a default position immediately preceding the verb.
The Latin examples that inform the authors' analysis of neutral word order are from the following three types:
a) overt Subjects, DO and IO in (di)transitive verbs (dare, donare, adimere, extorquere / eripere, praeficere, bellum inferre)
b) oblique arguments in verbs with DO + obliques (DO-Loc with conlocare, DO-Ablative of separation with liberare, DO-Instrumental (cingere, complere, cumulare, aspergere, oblinere)), DO-Goal phrases (in castra, in hiberna, legatos mittere, impetum facere), DO-Source phrases (ex urbe, ex arce, ex hibernis, ex agris, ex oppido);
c) Adjuncts (Instrumentals (gladio-ferro, causa), Time and place adverbials, Manner and means phrases (virtute, magnis itineribus, sine), Comitatives, Ablative absolutes.
Chapter 2 (pp. 145-224) deals with differences in the positions of the verb. The claim is that when the verb appears in the first position, this is mainly due to focus (section 2.1.). The evidence supporting the authors' analysis is a) polarity focus (yes-no questions, negative polarity), b) imperatives, c) thetic sentences (i.e., sentences presenting information as an occurrence rather than as a relationship with the verb), d) discourse cohesion operators (prosodic or encoded by particles) and e) conjoined structures or asyndetic lists of events (where theticity is also present for discourse cohesion). The verb in the second position usually occurs in sentences with negative quantifiers (nemo, nullus - ullus), which carry strong focus (section 2.2). The position of the auxiliary verb in analytic verbal forms ((amatum) est) in main and subordinate clauses is also discussed (including individual verbs in Caesar, mainly pugnatum est, interfectus est) (section 2.3). So is the copula (section 2.4), described as conditioned by factors similar to the auxiliary, and existential esse (section 2.5). The last part of this chapter (section 2.6) is devoted to a review of the position of auxiliary, copula and existential esse in Sallust and Cato.
Chapter 3 (pp. 225-313) deals with pragmatically structured sentences. Section 3.2 (Association with Focus) is a study of strong focus bound by lexical operators, that is, replacive negation (y non x, non x sed y, non solum sed (etiam) y), Yes-No questions with -ne cliticizing onto the focus element, frequency adverbs (semper, saepe, numquam) and focus particles (non solum / modo ... sed etiam, quoque, ne ... quidem) and chiasmus. In section 3.3 (Weak Pronouns) the authors discuss the WO of a number of pronouns grouped together under the term weak pronouns.
In my opinion, the two highlight Chapters are Chapter 4 (Arguments of Nominals, pp. 314-402) and chapter 5 (Modifiers, pp. 403-523).
Chapter 4 is devoted to nominal projections of verbal nouns (WO of subjective and objective genitives, partitive genitive and possessive genitive with relational nouns (such as filius, uxor)). Unlike the WO analyzed for sentences, a default WO cannot be proposed in NP, mainly because no clear generalizations surface. Section 4.1 deals with head nouns where a genitive can precede or follow the head depending on a number of factors pertaining to each different lexical item; the restrictions need to be explained by a set of properties (semantic, informational or phonological) and do not need to be directly related with the pragmatic side. A number of particular nouns are analyzed, representative of different semantic classes (eventive nouns with subjective and objective genitives, psych nouns, nomina actionis, nomina agentis, relational nouns), and also different types of genitives (possessive, partitive) in small samples taken from different authors, which provide preliminary orientation about the preferred (occurring more frequently) position of the genitive. The last section of this chapter (4.3) deals with Adjectival Phrases (using a data set of inmemor, expers, cupidus).
Chapter 5 studies attributive adjectives in continuous noun phrases. Just as with nouns, no default WO is proposed for the position of adjectives within the noun phrase, given the variation between preposition and postposition they show. Word order is sensitive to factors such as, first, the type of property denoted by the adjective (evaluation, measure, age, mental or physical state, material and color); second, their quality of absolute (e.g. suillus) or scalar (e.g. gravis); third, the syntactic hierarchy or scope (Quality > Size > Shape > Color); fourth, types of semantic combinations with the head (intersective, i.e. the result of a flat addition of the adjective property and the noun content, or intensional e.g. vetus, falsus, whose denotation depends on the properties of the noun). The analysis is applied to a number of adjectives of different classes (5.1 Intersectives, 5.2 Adjectives from Proper Names, 5.3 Age and Evaluation, 5.5. Measure; Staked Adjectives, and 5.6 Quantifiers; Demonstratives). Throughout the entire chapter there is an interesting comparison between adjective position in Cato and in Columella. The result reveals some diachronic changes. The last chapter, Hyperbaton (pp. 524-610), is devoted to hyperbaton (WO in discontinuous phrases, "perhaps the most distinctively alien feature of Latin WO" (p. 524)). This chapter deals with the discontinuity of genitive and noun (6.1), modifier hyperbaton when the modifier follows the head and when it precedes it (6.2 and 6.3), and other kinds of hyperbata. The analysis, as in other chapters, consists of searching the syntactic structures needed to account for the described types. Differently from the rest of the book, however, the analysis is presented together with the description. The book finishes here, without any conclusion, recapitulation or summarizing chapter.
This is a valuable study which for the first time presents a general perspective on the pragmatic role of word order in Latin within a Generative Grammar framework. The authors have achieved this by applying a method of analysis built upon a research program for many other languages. In my view this is the strongest but also the weakest point of the book.
The book's most valuable contributions are the large amount of data covered in the analysis and the depth of the analysis: despite the complexity of the subject, the authors do not omit any of the difficulties. Furthermore, I appreciate the authors' attempt to guide the reader through the difficult theory (see as an example pp. 25-29). The theoretical approach might alienate, I am afraid, some Latinists. Without going into the evaluation of this theoretical model, I believe that in spite of the authors' efforts towards clarity, the analysis will be almost impossible to follow by readers who are not familiar with research in GG. Personally I find so-called "scrambling" one of the weakest concepts of this theory, and for Latin this is crucial, because the reader gets the impression that it provides justification for WO that do not fit any other analysis.
From a philological point of view, the major problems are, in my opinion, the following. First, the selection of the corpus is problematic, mainly in chapters about WO in sentences. The conclusions would be stronger if the default WO of sentences had been based on the analysis of a fixed corpus of texts rather than in the selection of examples, which appear more as an illustration of the theory with Latin data than as an analysis of Latin itself. Contrary to the opinion of the authors, in a language with a closed corpus of texts such as Latin, drawing from a fixed set of data may be necessary. Indeed, the authors do present such numerical data -in small samples- as a preliminary orientation in other areas of this study. I suspect this is when an easy search with electronic tools was possible. For example, there is data for analysis of particular phrases (e.g., in castra, in hiberna in chapter 1, some nominal phrases in chapters 4-5, etc.). As for the interesting analysis of modifiers in Cato and Columella, they compare the number of cases of the orders head-modifier and modifier-head in both authors as a first look at diachronic trends. This procedure gives a more convincing initial approach to the analyzed phenomena, even if the samples are small and the results just orientative.
A second problem is that the authors offer data that cannot be properly evaluated from the contexts they present: in my view, many of the pragmatic phenomena proposed by the authors would have needed a larger context. This problem, which holds throughout the book, becomes particularly relevant in the analysis of nominal phrases: that a head or its nominal projection is presupposed or is focused information is not always immediately clear from isolated examples, which are limited in length, as the authors themselves recognize ("Pragmatic distinction can be quite subtle and open to conflicting interpretation", p. 36). The reader is left with the choice of believing it or not believing it. Naturally it is hard to avoid an impression of circularity.
Third, in terms of organization, a last chapter of conclusions, with a compilation of different factors conditioning the WO in Latin, would have been very useful. Partial conclusions are offered only in the Structural Analysis section of each chapter. Therefore, readers are left with the impression of many details that lack coherent organizing principles, especially for readers who skip the Structural Analysis sections.
Some philological aspects also require comments: First, no indications are given of the edition of Latin texts used for the examples. Second, references to Latin authors do not follow the standard form in Latin Philology (Thesaurus Linguae Latinae or Oxford Latin Dictionary.)
As for the bibliography, its purpose at the end of each chapter is not clear. The list of bibliographical references is biased towards the theoretical aspects of the GG model. That being the case, the fact that Chomsky is not mentioned at all (neither in the text nor in the references and bibliography) is rather puzzling.
Incidentally, I would add some minor formal points: The method of numbering examples is inconvenient for their later reference (see e.g. p. 243). A combination of numbers and letters would be easier to follow: (1a), (1b), and so forth. Additionally, Pesetsky 1995 (quoted on p. 138) is lacking in the bibliography, and Pinkster (quoted in p. 136) is lacking in the index nominum. A few Latin examples begin with uppercase letters (e.g. p. 449, ex. (156) Cato 69,1, p. 450, ex. (195), Col. 12,52,14-16, p. 455, ex. (176), Livy 40,42,1)), while most of them do not. Some Latin examples have a period (ending punctuation, presumably) before the parenthesis with the Latin author reference (see e.g. p. 473, ex. (234) Cic. fam. 11,21,2)), whereas most don't.
After reading the book, I have to admire the authors' courage in this huge undertaking. Their ambition in dealing in depth with Latin WO is both this book's main virtue and also defect. They use the Generative Grammar model, which could put off many Latinists, but the amount and variety of analyzed data make the effort worthwhile. Its reading is, in my view, worthy and recommendable.