1 Introduction

The syntax of Numeral-Noun Constructions, a construction in which there appears to be an apparent mismatch in case between numerals and the nouns that they quantify, is a point of contention cross-linguistically as there is no uniform way with which to account for the spread of data. The Numeral-Noun Construction itself is a construction in which a large amount of work has been done with respect to Slavic languages (Babby 1987; Franks 1994; Pesetsky 2012; Ionin & Matushansky 2018). In addition to these works, there is also a healthy and growing amount of research being done into similar phenomena in Finnic and Sami languages (Brattico 2011, 2012; Norris 2014, 2018; Nelson & Toivonen 2000). With that said, the analysis of such constructions is highly debated and inconclusive at best because of the highly variable nature of such constructions cross-linguistically. The book under review provides and showcases a detailed analysis in the study of the Numeral-Noun Construction (NNC) in Polish. Although deviating from standard analyses of the same construction done for Russian (such as Pesetsky 2012), the book under review offers a very cogent analysis of these constructions in Polish. The analysis in this book deals with data of the following type:

If their analysis proves to be correct, it may be possible to extrapolate their model to other languages within which NNCs of this type are observed, such as Finnish, Estonian, and the other Slavic languages.

2 Layout and analyses

The book begins with an introduction, continues with five chapters, and ends with a chapter summarizing the main conclusions. The introduction outlines the problem at hand and the authors' preliminary solutions to the syntax of the NNC. In traditional Slavic literature, this phenomenon is more familiarly known as the Genitive of Quantification (GoQ). It is so named due to the fact that the nouns that appear in this construction occur in the genitive case, instead of the more common nominative seen in many other languages. This process was initially most studied in Russian where not only the GoQ was present, but so too was an apparent semantic mismatch in the number of the quantified noun wherein the genitive singular appears with lower numerals. Data of the following type served as the basis of comparison for the Polish data analyzed in the book:

These data are only partially similar to the Polish data in question. In Russian, the numerals 2–4 are followed by the genitive singular in the masculine and neuter only—in the feminine, as in (3c), the nominative plural occurs after numerals. Finally, for numerals greater than or equal to 5, the nominal complement is in the genitive plural. ² Polish data for comparison are:

The examples in (4) showcase the following major points: i) numerals <5 occur with the nominative plural regardless of gender, a fact that stands in contrast to the Russian data; and ii) with numerals greater than or equal to 5, the genitive plural is the complement of the numeral and, like Russian, 3^rd person singular neuter agreement is invoked on the verb. Thus, as the authors point out, although there is a great deal of cross-linguistic similarities amongst the Slavic languages with respect to the NNC, they are in fact different and as such require individual scrutiny in order to map out their individual processes.

Further to this point are data of the following nature, which posit an additional issue unique to Polish, in which participles may agree with the genitive noun or may bear accusative case:

In order to account for the Polish patterns, the authors posit that NNCs of the type in (5) are actually in the accusative, not the nominative (the numeral itself is morphologically ambiguous, so the form of the demonstrative or the participle may be used to discern this fact). Secondly, the participle may appear in either the accusative, agreeing with the numeral, or in the genitive, agreeing with the quantified noun.

The authors begin deconstructing these data in the introduction by approaching them from a strongly Nanosyntax-based position. In particular, they posit that the appearance of an accusative numeral and genitive nominal complement in Polish NNCs is the result of a series of case projections (K_seq) which assign case in a semi-cyclic manner (cf. (15)). The Numeral-Noun Construction that the book discusses in particular, stands in contrast to a binominal structure in that it only has one nominal as opposed to two—that is, the authors consider the numeral to be a numeral and not a noun. In this analysis, the NNC occurs with only one instance of K_seq as opposed to two K_seq which each apply the required case individually in a binominal. They thus propose the following:

Chapter 1 is a cursory overview of the general data in Polish which has already been presented in this review. In this chapter, the authors demonstrate that the “low” numerals (2–4) behave like adjectives and the “high” numerals (5+) behave like a hybrid of both an adjective and a nominal. In addition, the high numerals also force genitive case onto their complements, in contrast to the low numerals. Of particular import in this chapter, are the following data:

On the basis of these data, the authors begin to substantiate an analysis in which the NNC does indeed occur in the accusative and not in the nominative, as masculine virile nouns in Polish show case syncretism between the genitive and accusative. In particular, the data in (7b) seem to show that this is indeed the case, as the nominative masculine virile demonstrative pronoun ci ‘these’ is ungrammatical in this position, and the genitive demonstrative pronoun tych ‘these’ is the only possibility in this case. It would have been helpful to more clearly delineate the difference here between the virile and non-virile numerals that are present in Polish, which might more clearly establish the oncoming Accusative Noun Subject (ANS) analysis. A further, clearer argument, is presented in the form of the accusative argument of certain impersonal verb constructions, in which accusative case is assigned by some other external head. The Polish NNC can occur with this construction in examples of the following type:

As these constructions must unambiguously occur with the accusative on the affected noun, the authors ascertain that the NNC must be an accusative and not some kind of aberrant nominative.

Chapter 2 goes into more detail about the GoQ which has been briefly mentioned in relation to Russian. The critical base that the authors draw from is of course Babby (1987), which essentially posits that Russian nouns in structural case positions will have their features copied onto other elements within their maximal projections. When a numeral is introduced, Babby posited the introduction of a full QP that took NP as its complement. This QP was argued to be the source of the genitive case on the complement noun and its modifiers. The noun however was the head of the entire structure. The original analysis by Babby (1987) and multiple analyses by Rappaport (2002, 2003) collectively seemed to point to an intimate AGREE relation between the numeral and the noun in which the numeral bears a valued case feature, and the noun bears a valued gender feature. A process of AGREE in both directions occurs and the numeral values the noun's case feature and the noun values a numeral's gender feature.

For the Polish data, this analysis falls short because of the fact that there is a difference in the way that virility affects the morphology of the numeral. Particularly, the morphological form of the masculine numeral is directly affected by the presence of the feature [±VIR] on the noun, therefore, a straightforward system of AGREE will not be enough to account for the data in Polish. The analysis of the Polish NNC is taken in parallel with Babby's (1987) original account that the noun is the head of the quantified phrase. The authors thus propose the opposite: when the numeral enters the derivation, it becomes the head of the construction rather than the noun. Evidence for this claim can be adduced from NP ellipsis in Polish:

According to this analysis, the numeral constitutes the head of a numeral phrase with the noun as its argument. The data in (9) are particularly interesting as they show a paradigm in which NumP is not elided in Polish. In these instances, only the noun is elided and NumP must necessarily be expressed. In order to account for this, the book discusses the Accusative Hypothesis, which is the claim that numerals are inherently accusative. Due to this fundamentally accusative nature, both lack of agreement with the verb and agreement of modifying demonstratives either with the accusative numeral or the genitive complement is observed. Crucially, Miechowicz-Mathiasen (2012) demonstrates a diachronic relationship between higher cardinals and accusative case. In her account, she posits the presence of a (null) functional head, p, which is the alleged source of accusative on higher numerals. This account relies on the morphological development of the masculine gender as well as the use of accusative case on numerals after other quantifiers. In Modern Polish, the numerals are argued to have become lexicalized and the NumP head merges with the aforementioned p. This p then morphologically merges with D and moves to spec, DP. With this established relationship, the interpretable features on p can check the uninterpretable features on D. In so doing, this allows D to become involved in probe-goal relations with external probes and consequently, in structural case positions, this allows the numeral to be marked accusative. This process can be summarized in the following trees:

This is followed by a m-merger of p and D (i.e. the aforementioned morphological merger):

The proposed light preposition p and its relation to D according to these structures results in the accusative. Miechowicz-Mathiasen argues, then, that p is defective, which makes it immune to outside case-checkers, but it does not mark D for deletion. As a result, a case other than accusative can be assigned to D, i.e. one of the oblique cases which will overwrite the accusative case resulting from the relationship between p and D.

The authors finish Chapter 2 with some other analyses regarding the Genitive of Quantification in which some numerals are Adjectives and some are Quantifiers. The major split between the two lies in the quantifiers' ability to assign genitive, and the adjectives' inability to do so.

Chapter 3 begins the main analysis of the book. The authors put forth the following conjecture:

This conjecture is a means of justifying the apparent matching of case in the oblique contexts, for example:

It is essentially a way to address the elsewhere cases of the NNC so that the authors can focus on the more interesting data—that is, the NNC in structural case positions. In these structural cases, the book offers two strategies that are in use in Polish: ³

In the analysis in (14a) we have a structure in which the numeral is in fact a head of an NP taking another NP as its complement. The structure in (14b) takes the alternative numeral-as-a-specifier approach which argues that the numeral is base generated in spec, NumP. Both analyses have their merits and as aforementioned, the authors of this book vouch for an analysis in which Polish executes both strategies. The first strategy is highly reminiscent of Norris (2018) in which a similar method is implemented for the Numeral-Noun Construction of Estonian—a different language, yet very similar in its expression of NNCs. ⁴ A cursory analysis of the cardinal as head analysis, i.e. (14a), for Polish returns negative results for the lower numerals as they function entirely as adjectives. In contrast, the cardinal as a specifier approach seems to be more a lucrative option for numerals in the book as it is readily able to explain the constituent-like nature that the cardinal numeral and any prepositions share. In the following examples, which demonstrate the latter claim, the preposition acts as an intensifier forming a PP with the numeral:

Arguably one of the most interesting parts of the book is the discussion of the development of the current NNC system in Modern Polish from Old Polish. Oftentimes, it seems as though the diachronic side of linguistics is overlooked, so it is refreshing to see the authors include a treatise of the history of the language as a source for the observed phenomena in Polish. One of the major motivations for the behavior of the higher numerals in Polish is the fact they were once nominals subject to the feminine declension (i-stem nouns) only to later develop into a hybrid of a noun and an adjective (which can be seen in their varied declension). A high numeral's nominal complement would then take the genitive case as just a regular process of the language. Indeed, this process is still seen in higher order numerals (e.g. tysiąc ‘a thousand,’ milion ‘a million,’ miliard ‘a billion’), which assign unmarked genitive case to their nominal complements—that is, they can be argued to be true binominal constructions. The current analysis of high numerals in Polish, according to Rutkowski (2007) is that these numerals turned into a functional Q category:

This structure provides some insight into the structural changes into a quantified expression. In addition, the authors posit that the accusative subject is in fact the result of the structural cases collapsing into one form (a process that is entirely plausible as it is extremely common, e.g. Ancient Greek to Modern Greek collapse of Dative and Genitive into one case).

As stated already, the book takes a Nanosyntactic approach to the structure of NNCs. Specifically, it makes use of the Universal Case Contiguity (Caha 2009, 2010), which involves a hierarchy of case projections, and also stipulates that in any given derivation, all case projections are present. The hierarchy is as follows:

This hierarchy forms the cusp of their analysis which will be used to assign the relevant cases in the NNC. In particular, the authors make a very cogent argument regarding a probe/goal relationship between the T_fin head and the NNC. For the structural instances of the NNC in Polish, T_fin probes deep into its domain until it comes across F_Q, the head of QP. It is at this point that the full QP moves into spec, AccP in order to receive accusative case. This movement involves pied-piping of the noun. The N head then enters into an agree relation with the Gen head in order to receive genitive case. Beyond that, the entire AccP moves into spec, vP in order to fall in line with the well-established verb internal subject hypothesis.

The authors, however, argue that at this point the rest of the case projections, i.e. instrumental, locative and dative are “peeled off.” Peeling off refers to the process by which the unused case endings are left stranded after movement has occurred (Caha 2009). In the event of such a movement, the case projections above the moved material (in this case GenP), are to be spelled out as some morphological material that may be phonologically null. Anything below the moved case projection is also moved but depending on which KP is being moved, the case ending may differ (e.g. if it is accusative, then the suffix on the whole phrase will be accusative). The difference in the Polish NNC is that movement of AccP to spec, vP does not occur, and instead, the AccP moves into GenP first so that the nominal complement can receive genitive case from the Gen head. The movement of the entire phrase from out of spec, GenP into spec, vP is conditioned by a second probe/goal relationship upon which v activates AccP on the entire QP, as should be expected. This entire process is contingent on a minimalist interpretation of Phase Theory in which the genitive noun being in the domain of F_Q is no longer accessible, however, the numeral, being in spec, QP, and on the left edge of the QP phase is open to further operations. In light of all of this, the process of the Polish NNC seems to be very well thought out. That said, the authors interestingly propose that, at all times, every single case projection is present in the domain of VP. In this manner, an appropriate k-selector projection (e.g. S-DatP, S-InstP, S-GenP, etc.) can cooccur with the relevant lexical case on a nominal argument that is base-generated within the case projection. Caha's (2009) original proposal, however, does not ascertain that this is the case in every projection. Caha's analysis was in the context of active/passive pairs that projected a specific case sequence: Inst>Dat>Gen>Acc>Nom. The projection of Instrumental was of course imperative because in passives, the instrumental marks the external argument. Crucially, however, this series of case projections appears low at PF but is base-generated in spec, vP. If a structural case is involved, it is subextracted from within the extended case projection and moved into a structural case position (e.g. spec, TP). In view of this, the structure that the authors propose does not fully demonstrate this analysis. It seems that the authors wish to indicate that the series of case projections is base-generated somewhere in the domain of VP. This does not reflect Caha's proposal. What we can instead assume is that only GenP, AccP and NomP are present as internal arguments in quantified structures—that is, we can forego the presence of InstP, LocP and DatP entirely for the analysis of Polish NNCs in structural case positions.

The next point of contention that the book sets out to explain is the grammaticality of either accusative or genitive modifying participles and demonstratives. In order to accomplish this the authors, citing Pesetsky & Torrego (2001), invoke the concept of equidistance. In short, the idea of equidistance stipulates that demonstratives and participles are base generated in a position that is equidistant to AccP and GenP, thereby allowing them to receive the appropriate case morphology from either one. They do also consider cases of the following type:

As can be seen, there is a case mismatch between the demonstrative and the participle. The book specifies that in these cases, the timing and point of adjunction of the demonstrative is variable, which would therefore lead to the mismatch. An analysis in which the demonstrative adjoins to the quantifier in GenP causes an issue of equidistance from both GenP and AccP. In so doing, the participle would be forced to appear as genitive, i.e. wybranych ‘selected.’ This would be pertinent to (19b), but as can be seen, this isn't the case. The authors propose that the case on the participle and the demonstrative is contingent on where exactly they enter the derivation. They suggest that in the case of the pre-quantifier, if it adjoins to Q_HP/Num_HP when QP has moved to AccP, it appears as accusative te ‘these’ thereby matching in case with the numeral. If the pre-quantifier is adjoined to NP, then it will take the genitive case upon movement into GenP, thereby agreeing with the noun and appearing as tych ‘these.’ In this adjunction, the demonstrative then takes the relevant case of the KP that it is a part of, i.e. matching with either the numeral or the noun. The adjunction of the demonstrative in this manner therefore has no effect on the participle, according to the authors.

Chapter 4 concludes the main analysis of the book and ties up a few loose ends that have not been addressed. It discusses “ramifications” of the key concepts discussed, however, beyond the first subsection of the chapter, there are no further effects of their analysis with regard to linguistic theory as a whole. That said, the first subsection discussing the ANS (Accusative Noun Subject) is very important to the presented theory and has theoretical consequences. This section deals with and accounts for the authors' evidence for the accusative subject in Polish NNCs. The book subjects the Polish NNC to a battery of subjecthood tests that indeed seem to unanimously indicate that the ANS is in fact a subject with some unique properties. For example, the Polish ANS occurs with default agreement, i.e. with 3^rd singular neuter agreement on the verb. This in and of itself is a property unique to the ANS in Polish. In sum, the book provides ample evidence in favor of an analysis in which the ANS is indeed a possible reflex of a subject and is equal to the nominative in every way except for in verbal agreement and in case.

The relationship between T_fin and AccP is an intimate one in the analysis of NNCs. In particular, the book underscores the fact that the probe on T_fin bears two prongs, one for case and one for phi-features. Of the two, the case feature is unvalued but interpretable and, following Béjar (2003), there exists a hierarchy between the structural cases, i.e. accusative over nominative case because the licensing conditions for nominative are “narrower.” Combining Béjar with their own hypotheses, the authors come to the conclusion that structural case is by default accusative and that nominative case overwrites it in syntactically salient contexts. The relationship between the T_fin head and the ANS is then twofold: it first enters into an AGREE relation with the noun, whereupon Agree-copy copies the phi-features and case features of the head noun. In the case of the ANS, however, the book proposes an extra caveat: the K_seq of the NP is inactive which thereby allows accusative case to be valued in T_fin as opposed to nominative. This analysis hinges on the fact that the ANS behaves and appears in the same contexts as a regular nominative subject. This in particular leads the authors to believe that the ANS is a simple exception. It is unclear, however, whether or not this property is because of the inactivity of the NP to enter into an agree relation with T_fin or whether there are some extant phi-features on F_Q which value T_fin first. This could be particularly interesting if one takes the numeral in these constructions to be nominal in its own right, which is not discussed by the book in the context of the ANS.

The treatment of complex numeral noun constructions provides a parallel analysis that, in truth, can be summarized in the context of a binominal configuration. It is no secret that complex numerals, more often than not (if not always), behave as nouns, so it should be no surprise if they appear to do so in Polish as well. However, the book offers a slightly different analysis. In complex numeral constructions, the book ascertains that these nouns do not have nominative case, and only ever appear in the accusative—in line with what has been established for higher numerals:

This can be contrasted with the following:

As can be seen, the verb agrees with the subject, which in this case is tysiące ‘thousands.’ If the construction with a lower numeral were used as in (18), then the expected NNC paradigm is observed, and any modifying participles will appear in the variable structure, e.g widziane/widzianych ‘seen.’ Interestingly enough, adding certain adverbs that affect the scope of the phrase, e.g. prawie ‘almost,’ will force agreement on both the auxiliary verb and the participle:

Without the adverb, the book notes that the default agreement on the verb is observed, e.g. było ‘was,’ and that the participle can optionally agree with either the QP or the NP in case. In complex numerals, the authors posit that their proposed process occurs between the higher numeral and the noun, i.e. between tysiące ‘thousand’ and studentek ‘students.’ With high numerals that also appear with genitive (5+) the following is observed:

The occurrence of two genitives posits two possibilities according to the book. One possibility is to add another QP₂, another is to say that genitive is an unmarked case after tysięcy ‘thousands.’ If this second route is indeed the case, it would not be farfetched to assume the binominal analysis of these nominal numerals and therefore stipulate that in all cases of these numerals, their nominal complements are simply in an unmarked adnominal case.

Chapter 5 approaches the content of this book from a minimalist perspective. Up to this point, the approach taken by the authors is a strongly Nanosyntactic one, and Minimalism has been peripheral at best. The cornerstone of this chapter is discussing the status of the NP as a phase, which, unfortunately is inconclusive in the literature (Bošković 2013; Citko 2014; Matushansky 2005). As the book points out, the NP on the PF side of the story acts as the quintessential “good” phase. It is open to a wide variety of movement-based operations and various procedures that the well-established vP and CP phases can undergo; however, the NP's LF behavior is opaquer. The authors—a s others have before—point out that the NP seemingly cannot be a proposition, i.e. <t>, and they are not open to QR in their left edges. In Slavic languages, it appears, however, that the latter of these issues is possible. This analysis is, however, at best tangential in the context of NNCs and not explicitly necessary to discuss in any real capacity. For the purposes of the NNC, a discussion of NP being a phase is pertinent only in the event that once NP is moved into GenP, it takes genitive case and then closes its phase, thereby immune to further operations.

Chapter 5 concludes with an experimental study conducted by the authors to study Polish Left-Branch Extraction (LBE) and essentially concludes that the presence of a QP renders LBE virtually impossible (or at the least very marked) since movement would have to occur through a phase edge, which could thereby violate the principle of Antilocality if this movement is too proximal. Overall, this chapter provides some insight into the minimalist interpretation of the main topic of this book, but was a touch tangential to their main argument.

3 Conclusion

This book, overall, provides an elaborate and well-laid out account of the Numeral-Noun Construction in Polish. This highly debated and rather complex structure is fairly complicated to map out and the authors have done a good job in doing so in such an accessible and easy to read manner. Their work could indeed provide a strong foundation upon which analysis of this same construction not only in Slavic languages but also in the Finno-Ugric languages may also be conducted. In this respect, this book could prove pivotal in bridging the gap between the way that NNCs work not only between these two language families, but also holistically. In that regard, this book has established itself as an invaluable source of material in shaping the way that we view the NNC. Its heavy use of Nanosyntax and Agree to achieve this goal could potentially be an issue to some, but that does not detract from its main arguments. Nonetheless, this book will prove useful to those interested in Numeral-Noun Constructions regardless of the language in question.