Hill and Mardale’s book The diachrony of differential object marking in Romanian stands out from the vast majority of works dedicated to the complex landscape of differential object marking in Romanian in several respects. First, as opposed to analyzing just the modern standard Romanian picture (as is commonly done), it focuses on the emergence and development of this phenomenon against a typological background. It thus combines two perspectives, namely diachronic syntax and comparative syntax. Second, it demonstrates that many aspects that have resisted explanation boil down to Romanian exhibiting a typological mix, which merges both Balkan and Romance patterns, each with its own traits, but nevertheless acting in tandem. And third, while the customary trend has been that of studying direct and indirect objects separately, the book examines the two in parallel, focusing on their commonalities and differences across all periods of Romanian.

Differential object marking (DOM) is a subclass of the wider, cross-linguistically robust phenomenon of differential marking by which certain types of nominals (animate, specific, definite, etc.) are flagged via dedicated morphosyntax distinguishing them from other nominals (Bossong Reference Bossong1985, Reference Bossong, Wanner and Kibbee1991, Reference Bossong and Feuillet1998; Comrie Reference Comrie1989; Aissen Reference Aissen2003; Seržant & Witzlack-Makarevich eds. Reference Seržant and Witzlack-Makarevich2018, a.o.). In modern standard Romanian, certain types of animates may require the presence of the prepositional marker pe, when used as direct objects. This marking, labeled DOM-p, presents non-trivial interactions with a doubling clitic (CD) in its accusative form, under the CD+DOM-p strategy. Differential indirect object marking, in turn, involves dative clitic doubling of a nominal which may carry overt dative inflection or be introduced by the directional preposition la ‘at’. Understanding the nature of these three strategies in various stages of Romanian is at the core of the book. An impressive wealth of data from the sixteenth to the twenty-first centuries, using a wide range of corpora, web searches, and constructed examples is illustrated and analyzed.

Besides Chapter 1, ‘Introduction’, and Chapter 8, ‘Conclusions’, the bulk of the discussion is contained in six chapters. With the exception of Chapter 6, ‘Formal approaches to DOM’, and Chapter 7, ‘A formal approach to Romanian DOM’, which contain formal accounts in the minimalist generative framework, the presentation is kept at a theory neutral empirical level, making the book accessible to researchers from various orientations. The book also contains an appendix on Romanian and Romania.

Chapter 2, ‘Two patterns for differential object marking: Balkan and Romance’, situates Romanian differential marking on direct and indirect objects in a larger cross-linguistic context, examining relevant structures in the Balkan Sprachbund and Romance. It starts by introducing the most important parameters in the description of Old and Modern Romanian DOM, taken to include indirect objects too; then the chapter underlines differences in the semantic triggers and morphosyntactic shape for DOM in the Balkan and Romance languages, leading to a better understanding of how these strategies blend in structuring the Romanian pattern. Generally, in Balkan Romance, that is, languages of the Latin phylum in the Balkans (Aromanian, Megleno-Romanian, etc.), the preferred DOM mechanism is clitic doubling with sensitivity to specificity, not animacy. Similarly, non-Romance languages of the Balkan Sprachbund use clitic doubling as a means to differentially mark both direct and indirect objects based on specificity, referentiality, and definiteness. By contrast, Old and Modern Romanian are similar to Romance languages in that animacy is a relevant parameter and a prepositional marker (DOM-p) is employed for differentially marking direct objects. But, at the same time, clitic doubling interacts with DOM-p, in ways that are neither fully Balkan nor fully Romance, as also seen from the discussion emphasizing the nature and the emergence of the DOM-p strategy in Romance languages as well as its microvariation.

Chapter 3, ‘Differential object marking in Old Romanian’, covers a period from the sixteenth century (when the first Romanian documents are attested) to the eighteenth century. It examines both direct and indirect objects, with exemplification of each of the three strategies: CD (indirect and direct objects), DOM-p (direct objects), CD+DOM-p (direct objects). The chapter examines the separate pragmatic effects of these mechanisms; in the period under analysis CD functioned as a common strategy for backgrounding (constructing a familiar topic reading), while DOM-p and CD+DOM-p were connected to foregrounding (indicating a prominent object, even if a familiar reading could still be maintained). Various important trends are illustrated: (i) although CD was possible with unmarked direct objects (unlike in Modern Romanian), it becomes the productive pattern with indirect objects; (ii) if in Modern Romanian CD+DOM-p is the preferred option, in Old Romanian DOM-p was instead quite robust with direct objects; however, a conjunctive type of differential object marking which encompasses both the animacy and the referentiality scales and which has persisted across all periods of Romanian is equally active; (iii) a third distinction: CD+DOM-p, which is the default option in Modern Romanian, is seen only with pronouns, and appears to have an emerging status.

In general, an unstable system for differentially marking direct objects is at work over this period: various categories that would need obligatory DOM-p in Modern Romanian can be found unmarked, or even clitic doubled in the absence of DOM-p, and instantiations of p(r)e as a generalized accusative case marker, oblivious to semantic scales are attested. A telling example is the grammar of Dimitrie Eustatievici Braşoveanul, printed in 1757, where the prepositional marker extends to all direct objects, irrespective of animacy/specificity (pp.52–57). The chapter also identifies important differences between Old Romanian DOM and Old Church Slavonic DOM, strengthening the conclusion that (Old) Romanian does not use a type of differential object marking as seen with the genitive in the Slavic languages.

Chapter 4, ‘Differential object marking in Modern Romanian’, is dedicated to a comprehensive overview of the Modern Romanian picture, which covers a period from the nineteenth to twenty-first centuries. The non-trivial complexity of the various differential marking strategies is emphasized and important distinctions from Old Romanian are outlined: the decay of CD with unmarked direct objects, the spread of CD+DOM-p as the regular option for marked objects, and strengthening of CD with indirect objects. The spread of clitic doubling with direct and indirect objects results from the strengthening of the pronominal clitic system in the language. Dative clitic doubling extends and consolidates with dative inflected or la-marked indirect objects. On direct objects, accusative clitic doubling is lost with unmarked nominals, becoming standard instead with the pe-marked ones.

On the pragmatic side, the foregrounding effect of DOM-p characterizing Old Romanian disappears. DOM-p is now a means to look forward in the discourse, while CD+DOM-p (discussed in detail in Section 4.3) can look back (i.e. via referential distance) and forward (via referential persistence). As a clear trend, differential marking is no more a salience/foregrounding mechanism for the object, but only signals referential persistence in the continuation discourse. This grounds the shift towards semantic, syntactic, and grammatical specifications as opposed to discourse foregrounding for differential marking. An important part of the chapter examines crucial differences between clitic doubling and clitic left dislocation. Despite homophony in the surface form of the clitic, the two processes have distinct natures, are subject to distinct semantic restrictions (for example, animacy is overridden in clitic left dislocation, etc.), and diverging trajectories. Collapsing both under the same umbrella has introduced unwanted noise, obviating their true nature; instead, the two processes must be studied separately, each in its own right.

Chapter 5, ‘The grammaticalization of pe’, summarizes the diachronic view on the pe marker and assesses the most important analyses and theories with respect to its nature. In Romanian this morpheme is homophonous with a locative preposition, meaning ‘on’, and descends from the Latin preposition per that carried polysemous spatial, temporal, instrumental, or manner content. Repurposing for differentially marking direct objects is hypothesized to have evolved from semantic bleaching, the loss of case assignment capacity, and reanalysis. In Old Romanian, the result of these processes is the construction of a topic marker, for example, by extension of the ‘concerning/aboutness’ meaning to DOM when the speaker’s intention is to place the direct object in the spotlight (Onea & Mardale Reference Onea and Mardale2020). The Modern Romanian outcome is a marker for animacy and specificity. Hill & Mardale also emphasize the important semantic and syntactic differences between the use of pe as full-fledged preposition (impossibility of clitic resumption or doubling, no passivization, etc.) and that of differential marking, which gives the opposite results on all tests. Another important conclusion is supported, namely that DOM-p in Romanian does not entail only bleaching of circumstantial semantics, but a more radical syntactic process which involves a reanalysis from a preposition to an element of the nominal domain. A detailed comparison with differential object marking in Spanish and Sardinian strengthens this observation.

Chapter 6, ‘Formal approaches to DOM’, initiates the formal part of the book, continued in the following chapter. It reviews formal accounts formulated for Modern Romanian DOM, which have generally been extended from proposals for other languages. Their insights, but also their problems in offering a unified view of the phenomenon are presented. Two crucial observations are supported; first, the analyses put forward for Modern Romanian cannot be easily extended to Old Romanian; second, both Old and Modern Romanian data are problematic for accounts in terms of raising or in terms of Kayne’s Generalization (which establishes a tight connection between the prepositional marker and clitic doubling under the prediction that the latter should be obligatory; see especially Jaeggli Reference Jaeggli1982). Various challenges to formal views that derive pe as a TP/vP element for Romanian can be alleviated under the hypothesis of a nominal nature of this marker – that is, seeing pe as an element merged in the extended nominal projection. With respect to the doubling clitics, both Old Romanian and Modern Romanian give evidence that they function as agreement markers (as in other Balkan languages) and not as thematic markers (as more generally across Romance). The nominal nature of pe can explain these interactions too.

Chapter 7, ‘A formal approach to Romanian DOM’, contains the formal account couched in a cartographic perspective. The gist of the analysis is that important results can be obtained if the nominal domain (as opposed to the verbal domain as in Spanish) is explored in detail for Romanian when it comes to differential object marking, such as to reveal the contribution of nominal mapping features which override verbal parameters. Under the proposed analysis, both the prepositional marker and the clitic undergo first merge within DP. Romanian verbs license all direct objects (marked or unmarked) in the same way – theta role and case are checked on first merge. Marked objects might have a reflex on the clausal spine as a side effect of discourse features. The chapter also addresses a phenomenon characteristic to Romanian by which overt definiteness results in ungrammaticality on unmodified direct object nouns under prepositions, diverging from the common trend to treat it as a PF mechanism. The clash is instead taken to result from the exclusion of the [def] feature (and subsequently of the DP field) from the relevant configuration, as a result of head restructuring. Through a detailed cartographic presentation of DP structure in Romanian (Section 7.3), the pe marker is assumed to merge within the DP as a D functional element. It is shown that a better understanding of crucial features which are generally less explored in the extended nominal periphery is needed in order to get an adequate picture of what differential marking indicates on direct and indirect objects. Among these are particularization ([particularize]) and speaker’s point of view ([Fmark]). For example, on direct objects, the main function of pe as a D element is to check the [particularize] feature. However, as shown in Section 7.4, the clitic takes over the functions of pe when the latter undergoes semantic beaching. This leads, among others, to a compromise between the Balkan pattern (clitic doubling for differential object marking) and the Romance pattern (the use of a preposition). The contribution of the two features when it comes to CD on indirect objects is also explored in detail.

To conclude, the richness of the data, the typological perspective, and the thorough diachronic investigation make The diachrony of differential object marking in Romanian not only a welcome contribution but also an essential tool for researchers interested in the nature of differential object marking, clitic doubling and clitic dislocation, and, more generally, Balkan and Romance linguistics, or diachronic and typological syntax.