The Oxford Latin Syntax (hereafter OLS) is a major contribution to our understanding of the workings of the Latin language. P.'s approach to Latin syntax differs markedly from such notable previous grammars as that of R. Kühner and C. Stegmann (Ausführliche Grammatik der lateinischen Sprache. Zweiter Teil: Satzlehre [1912–14]), to which P. explicitly sees OLS as a successor, but OLS is destined to be a similarly crucial reference work.

P. sets out to describe the syntax of Latin as represented in the surviving evidence of the language from the period c. 200 bce to c. 450 ce, taking due account of variation and change in the language within these bounds. His goal is to ‘define the rules underlying the utterances that are actually attested’ (I, p. 7) and to do so with attention to the fact that these utterances are produced in contexts and situations as instances of human communication. These two aspects, concern not to neglect diversity within Latin and to view each piece of evidence as the product of human interaction, have long been important in the study of Latin grammar. However, P. builds into the structure of his presentation the fact that particular usages both sit alongside the rest of their text and contrast with (at least) other ways that the same or similar things could have been expressed by the same writer or by others. Thus, chapters and sections deal not only with units of form that can be recognised purely grammatically, such as noun phrases or finite verbs, but also ones that are recognised by their functions, such as expressions of manner. For P. the connection between linguistic structure (selection, combination and ordering of lexical forms, inflected as required) and function (both semantic and pragmatic) is crucial, with each used to shed light on the other.

The first volume's thirteen chapters concentrate on simple clauses and their constituents. After an introduction (1) briefly defining the scope and aims of the work, two chapters set out notions that are basic or fundamental to P.'s treatment (2: constituent types; 3: word classes). P.'s overarching analysis breaks clauses into a core or nucleus (consisting of a verb and its arguments) and satellites. The central importance of verbs to the analysis of clauses is followed up in two chapters that address valency of verbs (4) and voice (5). There are then three chapters that broadly concern the clause as a whole, on sentence type and force (6), the semantic values of the tenses and moods of verbs (7) and negation (8). Arguments (9) and satellites (10) are examined in the next two chapters before chapters on the noun phrase (11, incorporating discussion of adjective phrases) and on the case and preposition system (12). The final chapter considers agreement within verb phrases and noun phrases (13).

The second volume, concentrating on how Latin clauses may be connected to one another, continues the chapter numbering and begins with five chapters concerning subordinate clauses: they are considered first in general (14) and then classified as functioning with verbs as arguments (15) and as satellites (16) or with nouns or other units (17); relative clauses form the focus of the final chapter in this group (18). Relationships other than subordination are addressed next in chapters on coordination (19), comparison (20) and secondary predication (21). Finally, P. looks at the representation of information structure (e.g. new vs old, topic, focus, contrast etc.) within clauses (22) and across connected discourse (24), with specific attention given to word order (23).

A comprehensive treatment of a language's grammar is of necessity a lengthy work of reference, and relatively few readers, as P. acknowledges (I, p. 8), will read OLS in its entirety in the order in which the material is presented. Users will certainly be grateful for the fact that the second volume contains a consolidated index to both volumes (as well as an index locorum) and that each volume begins with a full section-by-section contents list. Between these two aids, both readers who are familiar with the structure and terminology of more traditional grammars centred on word and clause classes and those who have become familiar with P.'s approach based more on function should be able to navigate the 2,531 pages of main text to find appropriate sections with reasonable efficiency. The hierarchical structure of presentation will suggest to users the relevant wider analytical or descriptive context for particular sections, while P. also provides plenty of cross-references.

Although such connections and dependencies across sections are integral to this kind of work, each section is sufficiently complete in itself to be useful to readers searching for information on a specific point. P.'s presentation in each section, for example 8.39 ‘Coordination of a negative imperative clause with a preceding positive imperative clause’, is consistent and clear. Introductory comments outline the basic pattern under consideration together with any important variations or alternatives and observations on their distribution; these are followed by illustrative examples for all the patterns described (accompanied by translations), frequently followed by a smaller print supplement of further (untranslated) examples; some sections end with a small print section of supplementary analysis or comment. The layout of the core examples, in which each begins on a new line, and the fact that even supplementary examples are also quoted in full has doubtlessly increased the page count but is greatly to be commended.

The broad textual basis that underpins OLS is crucial. The exemplification, drawn to a large extent from communicative texts such as the plays of Plautus and speeches of Cicero, is a constant reminder of the primacy of spoken language in P.'s thinking. However, other authors and text types are by no means excluded either from the analysis or the choice of examples, and inscriptional material is adduced when relevant. The particular prominence of Plautus as a source for quoted examples may give an appearance that OLS is somewhat divorced from the language found when later the classical standard emerges and holds sway in the usage of many texts with which readers of OLS will be concerned. However, there is arguably more continuity than change in Latin syntax through the period covered by OLS, and P. is careful to remark explicitly and give other examples when Plautus, Cicero or any source would present an incomplete or misleading picture. Thus, alongside patterns found widely throughout the evidence, P. notes plenty that are found only in some period, authors or text types (e.g. the use of historic infinitives in subordinate clauses in some historians at I, p. 608, or adnominal accusative case direct objects with -tio nouns in Plautus at I, p. 1041).

P.'s thorough examination of the corpus, aided by resources such as the TLL and electronic databases unavailable to most previous scholars, means that there is no significant area of Latin syntax without due attention. There is full treatment of the uses of the moods, the construction of conditional expressions, indirect speech and all the familiar topics of Latin syntax. Many discussions are supported by statistical evidence of distribution or relative frequency in different parts of the corpus (e.g. I, pp. 1180–3 charts relating to the distribution of cases and prepositions, II, p. 60 a chart recording the distribution of quod- and quia-clauses in Cicero's orations). Although it can be hard in places to know quite what to make of this information – a structure is either grammatical or not in a variety of a language, and frequency effects may be the result of many chance factors –, it is helpful as a reminder that often not all alternatives are equal: frequency of forms or constructions in differing contexts is part of a user's knowledge of their language.

Even rare phenomena can be tracked down in OLS, albeit sometimes only after some searching. Thus, the emphasising or clarifying ‘at that’ use of is (quod multi, et ii docti, fecerunt ‘that which many men have done, and learned ones at that [or and they learned ones]’, Cic. Sen. 84) appears under ‘epitactic coordination’ (II, pp. 693–6). The rather uncommon (and possibly Graecising) apparent nominative and infinitive of indirect speech (phaselus ille quem videtis, hospites, | ait fuisse navium celerrimus ‘My friends, that ship which you see says she was the swiftest of ships’, Catull. 4.1–2) is considered an instance of the prolative infinitive (II, pp. 168–9, so indexed, but cross-references would have been helpful from the two sections headed ‘the nominative and infinitive construction’ at II, p. 20 and II, pp. 194–202, since P. acknowledges this usage has been so labelled by others). Among possible gaps, the treatment of numerals, especially the ordinals, distributives, adverbs and their internal structure, is limited, but the evidence may be insufficient for a fuller account; the same may be true of the ordering of persons within coordinated phrases (tu et Tullia, ego et Cicero etc.), which is given only the briefest of mentions (II, p. 958, not indexed under ego or tu).

Like previous works, OLS covers some areas that might not meet a strict modern linguistic definition of syntax, including, for instance, the semantic values of the tenses in the first volume, the treatment of particular lexical items in both volumes (e.g. quis, idem, ipse in the first volume and imprimis and iam in the second), and the discussion of information structure and discourse cohesion in the second. These inclusions, however, are a strength: they situate the work in the broader study of Latin grammar, and they have found a place in previous works in this area; P. is generally careful to highlight their syntactic significance proper (e.g. their restriction to certain grammatical contexts, their effect on the ordering of elements); and P. deals with matters such as information structure and discourse in a more substantial way than previous works.

The major difference between OLS on the one hand and Kühner and Stegmann and other previous works on the other hand is without doubt the core, nucleus, satellite framework, and this is also one example of P.'s use of terminology that may be unfamiliar to much of his readership. Not all is unfamiliar, so P. still uses such terms as ‘ablative absolute’ or ‘purpose clause’. However, it will repay the reader's effort to stick with that which is not familiar. Terms are always defined, and P. does not use them capriciously or for their own sake; rather they emerge from his perspective on the evidence. The framework of P.'s approach is set out at some length in the opening chapters as well as at the start of individual chapters and sections. These discussions should not be overlooked for their fresh insights.

OLS's terminology does raise the occasional further challenge. Sometimes this is because the term is ambiguous: although P. mostly labels units by function rather than form, it might still be clearer to reserve ‘imperative’ for the mood of the verb (i.e. form) and not use it of clauses with reference to their function, especially not indirect commands (e.g. II, p. 126 ‘[subordinate] finite imperative clauses’). Elsewhere it is because P. uses terms in a different way from others, either in Latin linguistics or more generally, as, for instance, ‘topic’, ‘focus’ and ‘theme’ in the second volume: P. does define his use of these terms, and they are in any case ones that lack consistent use in the literature, but in a work that can be expected to be consulted piecemeal they are potentially problematic.

Throughout the analysis P. supplies helpful references to other scholarship on the same or related topics or material, offering further data, complementary or alternative approaches etc. Some relate to textual difficulties that P. is at pains not to conceal or minimise. The resulting extensive bibliography is a major contribution in its own right.

The presentation of the text is impressively accurate throughout: misprints and typographic errors are very few. One inconsistency between two quotations of the same example did present itself (I, p. 315 Ai'n tu, geminos? vs I, p. 323 Ai'n tu? Geminos?, Pl. Am. 1089), but in general the texts are presented as expected from their sources (with due attention to textual difficulties).

One presentational feature calls for some comment. P. has sensibly employed a hierarchical structure that can be observed in the contents pages and, in principle, the various typographically distinct levels of section headings. However, the section numbering is continuous within each chapter, regardless of the status of the section within the hierarchy. In that P. has subdivided freely down to as many as six levels in places, readers may be grateful for the brevity of 4.64 over, say, 4.5.1.3.1b as a cross-reference. However, the continuous numbering is rather unsatisfactory. Nuanced typographic distinctions between heading styles as the sole marker of the hierarchy of presentation are less helpful on the page than explicit numbering would have been, and cross-references to superordinate sections are inherently ambiguous as to whether they refer readers just to the section so numbered or also include any subordinate sections it may have (e.g. 22.25 ‘Quidem and equidem’ is superordinate to 22.26 ‘Quidem’ and 22.27 ‘Equidem’). The running heads are also frustrating: section numbers are absent from running heads in the second volume altogether, and in the first, though present, they are often unhelpful in corresponding only in a loose way to the actual number of the section(s) on the page. The recto running head on I, p. 355 is ‘6.22 Interrogative sentences (indirect illocutionary force)’, the current second-level subdivision, but the text on the opening is in 6.29, which begins on I, p. 350 and continues to I, p. 356; 6.22 proper appears on I. pp. 342–3. Since cross-reference throughout the text (but not the index) is by section numbers, ready access to the current section number(s) on every opening would have been helpful.

There is no easy way to divide a work of this kind, but many of the chapters are very long and/or have dozens of subsections, for example Chapter 7 (292 pages, 164 sections), Chapter 11 (242 pages, 160 sections). Shorter chapters (roughly corresponding to current first-level divisions) grouped into parts (corresponding to current chapters) might have allowed for explicit hierarchical section references without these becoming overlong: thus, for instance, the 98 pages of section 15.3 ‘Finite argument clauses’ and its subordinates (down to 15.89) might have made a chapter in itself.

Overall, the OLS is a monumental achievement, and such idiosyncrasies as there are do not detract from this. These volumes present a comprehensive unified account of the whole of Latin syntax, broadly conceived, and they are both a source of reference for the familiar and also an invitation to think about it anew.