This is the first book-length work devoted to a topic with a long tradition in linguistic research, specifically the Evaluative Binominal Noun Phrase (EBNP). The book examines the construction as a member of the English of-binominal family and situates it in this network, expanding on previous research. In particular, it discriminates different EBNP constructions which have been subsumed under the same heading in previous studies, namely, the classic EBNP, the evaluative modifier (EM) and the binominal intensifier (BI) constructions.

The book starts with an introductory chapter where the main properties of binominals are sketched and the scope of the study is delimited. The deviation from the rules characterizing canonical of-phrases, their creativity, their pervasive use in all English varieties and in many European languages as well as their long-standing use, which can be traced back to the thirteenth century, provide a plethora of ingredients to make it worth the exploration. The core of the book is divided into three parts, which provide a comprehensive description of the EBNP (part 1), a corpus-based analysis where selected data are tested against the previously described criteria (part 2), and a theoretical account from the perspective of two current competing linguistic models (part 3).

Chapter 2 provides an overview of diagnostic tests used in previous studies to classify English of-binomials. These tests aim to determine the head status of the first or second noun, constituency, constraints on the two nouns, selection restrictions on determiners and the status of the preposition. Headedness is examined through semantic, morphosyntactic and pragmatic discourse criteria. Constituency is tested using classic constituency tests such as preposing, postposing and coordination of the NP. Constraints on the two nouns help determine their degree of nounhood, their relationship and the function of the binomial. Determiner selection reflects changes in noun status or the construction's function. Finally, the evolution of the preposition of sheds light on the categorial distinction of of-binomials, ranging from conveying possession to serving as a linking device. The chapter concludes with a description of the methods and data used for classification, primarily sourced from the Corpus of Contemporary American English (COCA) and the Corpus of Historical American English (COHA), which means that they are not completely updated, although the data are recent, dating back to 2019.

Chapters 3 and 4 of the book extensively apply the previous tests to various binominal constructions, aiming to distinguish between them and to argue for the categorization of different types. Chapter 3 explores three constructions that mark the initial stages of grammaticalization from typical N+PP structures (e.g. any dog of respectable character) to pseudo-partitives (e.g. a sea of people). In between these two constructions are head-classifiers (e.g. wild flowers of spring), which have often been overlooked in previous studies but are thoroughly examined in this book. Unlike typical measure nouns, the first nouns analyzed here often lend themselves to evaluative interpretations. The chapter compares and contrasts these constructions in detail, highlighting their distinctive properties. In summary, the three constructions are differentiated as follows: in N+PP structures, the first noun is the head, the preposition often retains semantic content and the second noun functions within an NP embedded in a prepositional phrase. Head-classifier constructions classify or attribute taxonomic properties to the referent of the first noun, imposing restrictions on the second noun, which is non-specific and non-referential. Pseudo-partitives involve a shift in headship from the first to the second noun, with the first noun indicating quantity, measure or shape of the second, which must be a mass noun or bare plural. This book's contribution lies in elaborating on the connection between head-classifier and pseudo-partitive constructions, which expands upon existing research focusing on measure-first nouns.

Chapter 4 addresses the more subjective construction in the network of of-binominals in detail, tagged as evaluative of-binominals, and argues for a division into three separate constructions, i.e. the extensively discussed EBNP and two other categories. The different diagnostic tests introduced in the preceding chapters are now applied so as to draw a distinction between the three categories. Broadly speaking, the EBNP is an of-binominal construction in which the first noun has lost most of its features of nounhood and ascribes a physical (or abstract) property to the second noun, or where both nouns are compared (e.g. his shiny egg of a head). Because of this modifying function, the first nouns in this construction can usually be replaced with a corresponding adjectival modifier. The second noun, for its part, denotes the referent of the whole construction. As regards the other two constructions, the first one, referred to as evaluative modifiers (EM), has been completely ignored in previous research. In EM the first noun's ascription of a physical property to the referent of the second noun has been lost (e.g. some dragon of a receptionist). Instead, the first noun expresses a bounded positive or negative extreme on a scale and the underlying subjective-evaluation semantics of the noun is foregrounded. Structurally, the first noun has lost its nounhood, while of and the second determiner no longer have a clear syntactic function. Together, the first noun, of and the second determiner make a single constituent. The remaining third category has only been mentioned peripherally previously and is labelled as binominal intensifiers (BI). In BI all nominal properties of the first noun as well as its semantic content have vanished and the noun functions as an intensifier (e.g. a goddamm bitch of an unsatisfactory situation). By tracing the development of the three constructions, ten Wolde convincingly shows the path of grammaticalization of EBNPs. The contribution of the findings in this chapter is twofold. On one hand, EM has, for the first time, been identified as a separate category situated on the path to grammaticalization between EBNP and BI. On the other hand, ten Wolde observes that BI is much more productive than anticipated in previous research, which had been restricted to sort of, kind of and type of nouns and to quantity/size first nouns (a lot of, load of, a bunch of).

Chapter 5 explores the diachronic development of three nouns that have been carefully selected for being a frequent conventionalized pseudo-partitive (cake), a frequent first noun used in EBNP and EM (beast), and the most frequent first noun in EM and BI in PDE (hell). Furthermore, the study examines whether this transition can be classified as grammaticalization or lexicalization. Specifically, ten Wolde seeks evidence of decategorization, semantic bleaching and subjectivization – factors that collectively characterize a process as grammaticalization. The author concludes that the development in the three case studies exhibits properties indicative of grammaticalization, in spite of the fact that the ultimate end of the historical process has resulted in new lexical items (helluva, hella, beasta). First, as regards decategorization, the first noun loses its nounhood and becomes an adjective, bringing about a shift in the construction head and an internal reorganization of the phrase. Secondly, semantic bleaching is evinced by the shift from a lexical first noun evoking a referent to a lexical noun ascribing properties to the second noun to a grammatical intensifier. Finally, subjectivization is seen in the subjective, speaker-oriented evaluation required by the EBNP, which is foregrounded in the EM construction.

Chapter 6 uses modern corpus data to investigate the constructions discussed in chapters 3–5, focusing on the development from N+PP to head-classifier, pseudo-partitive and EBNP constructions. It analyzes the first nouns nub, breeze, husk and snake to explore evidence for this evolution. The chapter also considers alternative roles of the pseudo-partitive in this development. Later stages of this evolution, involving EBNP, EM and BI constructions, are examined using the first nouns whale and bitch. The proposed grammaticalization path begins with the reconceptualization of the relationship between the two nouns in the N+PP structure, where the second noun transitions from indicating properties like location or possession to denoting a defining property of the first noun. This transition is triggered by the loss of semantic content in the preposition of, leading to the head-classifier construction. In the pseudo-partitive construction, the second noun becomes the head and the first noun functions as a complex quantifier, simplifying the original N-of-N structure into a simple NP. The transition from pseudo-partitives to EBNP is less clear-cut. Two scenarios are proposed: a direct link between the two constructions or an indirect link, where pseudo-partitives lead to EBNP via different paths depending on whether the first noun is animate or inanimate. The development from EBNP to EM and BI appears to occur directly, with expansions in the second noun's collocates and increased frequency. Overall, the chapter provides valuable insights into the development of these constructions. Transitions from N+PP to head-classifier to pseudo-partitives are well explained, but the role of pseudo-partitives in the transition to EBNP remains uncertain due to semantic ambiguities and fine-grained distinctions between the constructions.

Chapter 7 investigates the development of premodification patterns in the various of-binominals discussed earlier in the book to ascertain whether these patterns offer evidence for distinguishing between the constructions and for the proposed grammaticalization path. The chapter also aims to explain the irregular premodification pattern observed in the EBNP, where modifiers of the first noun could have been selected by the second noun (her big lovable bear of a husband).

The chapter begins by explaining the coding system used for premodifiers, based on Ghesquière's (Reference Ghesquière2014) Cognitive-Functional Model, which categorizes premodification into zones such as categorization, modification and determination. This function-based approach helps disambiguate the roles of premodifiers and is suitable for studying the development of constructions as premodifiers tend to shift from right to left during grammaticalization. The chapter then conducts two corpus-based analyses of premodification patterns in the different constructions, examining the distribution between premodification before the first and second nouns and exploring characteristic premodifier zones/types for each construction.

The findings indicate that, although each construction has characteristic premodification patterns, the first three constructions do not significantly differ in premodification distribution, except for a preference for premodification before the first noun in head-classifier constructions. Evaluative constructions, however, show sharper distinctions, with EBNP favoring premodification before the first noun and EM and BI favoring premodification before the second noun. On the whole, the findings support distinguishing between evaluative constructions and suggest that EM and BI involve the conflation of separate premodifier zones into a single NP. The shift in premodification from the second to the first noun in EBNP marks the beginning of this process. Consequently, the value of this book goes beyond its contribution to the longstanding discussion of EBNP and expands to the research on premodification by shedding light on the development of of-binominals from descriptive to intensifying functions.

Chapter 8 offers an account of the evaluative of-binominal family network using the theory of Construction Grammar (CG). The chapter starts with an introduction to the theory's basic notions of construction and constructicon, followed by an explanation of the method by which constructions change and new constructions emerge via usage-patterns. Ten Wolde then adopts the constructional approach to model the different constructions. She builds the different formal and semantic properties characterizing each construction, mapping out a detailed description of each individual node in this network. In addition, ten Wolde relies on the empirical studies in part II to propose a scenario in which these constructions developed using the CG framework.

In the author's view, the grammaticalization process started with the compositional N+PP construction, where the embedded prepositional phrase ascribes a property to the first noun. The N+PP temporal construction (e.g. the breeze of the morning) would be on the same level as other realizations of the N+PP such as origin or location (e.g. the beast of Prague), which would constitute different nodes joined by the higher-level ‘mother node’, that is, the more schematic N+of-PP construction. Subsequently, at some point on the construct level, the PP is reconceptualized and, according to the author, metaphorically extended, denoting the first noun's class instead of ascribing a property to it (e.g. a breeze of morning). In this new head-classifier construction there is also formal reanalysis so that the PP is no longer a separate constituent, of becomes a linking device and the second noun is not referential, which explains that the second determiner slot is now empty.

The head-classifier construction is connected to classifier premodifiers (e.g. morning breeze) via a polysemy link in the constructicon. Ten Wolde suggests that it is this connection that might have created a network link between head-classifiers and the noun phrase construction. Subsequent changes concerning the reduction of the complex N+PP to a simple NP are the consequence of that connection.

Another node in this network is formed by the EBNP, which is completely different from the N+PP and the head-classifier both functionally and formally. Semantically, it is now the first noun that ascribes a property to the second, which denotes the referent and serves as semantic head. Additionally, while the second noun slot is almost unconstrained, the choice of the first noun is restricted. Formally, the second noun appears to be the syntactic head, both first and second determiners are restricted and premodification patterns are idiosyncratic. Again, metaphoric associations connect the EBNP to the head-classifier. Thus, while beast denotes an animal in the head-classifier, in the EBNP it ascribes animal properties to humans first and then to objects. This construction is linked to the most schematic N+PP construction, which explains that the indefinite article appears again in the second determiner slot. The EBNP is also connected to the simple noun phrase. Specifically, it functions like descriptive premodifiers (e.g. a hellish hotel), since the first noun ascribes an evaluative property to the second. Ten Wolde contends that this change in the function of the first noun is the initial stage of the subsequent reanalysis of the [N of a] modifier phrase. She argues that this connection to the simple noun phrase also explains why evaluative of-binominals follow the grammaticalization path of prototypical premodifiers, which shift from objective descriptive to subjective descriptive modifiers and finally intensifiers.

The EM construction is very similar to the EBNP, although there are important semantic and formal differences. Semantically, the first noun conveys the speaker's evaluation of the referent. Morphosyntactically, the first noun is restricted to singular number, the second determiner does not mark number and the two nouns do not necessarily agree in number. Finally, the most frequent first nouns are orthographically amalgamated with the preposition and following determiner (e.g. helluva), and sometimes reduced (e.g. hella). Ten Wolde observes that the transition from the EBNP to the EM is very often based on a metaphoric association, whereby the large size of the referent is mapped onto the speaker's positive evaluation (e.g. a whale of a movie). In her view, the EM is also linked to the simple noun phrase, with which it shares more formal features than it does with the of-binominal.

Lastly, in the BI, the [N of (a)] chunk has become an integral part of the modifier phrase and serves as an intensifier of the following adjective (e.g. a hell of a large sum). In this construction, the first determiner has scope over the whole construction, while the second one has lost all functionality. Unlike the EM, premodification of the first noun is restricted and the [N of (a)] chunk can sometimes be used predicatively. The link between the BI and the EM appears on the micro-construction level through common first nouns in partially schematic constructions, but in the BI [N of (a)] has integrated into the AdjP. The autonomy of this chunk when used predicatively appears to indicate that the BI is connected to the simple adjective phrase. Only on a more abstract level is there a connection between the AP and the NP. Finally, in the BI the connection to the schematic of-binominal has weakened.

Crucially, chapter 8 shows that the theory of CG is well equipped to explain the changes that the first nouns undergo along the grammaticalization path, and to predict how this path may have been created. However, ten Wolde objects that this theory cannot explain the order of internal changes found in the data, among other features. This gap is covered by the second theory discussed in chapter 9.

Chapter 9 presents the analysis of the different binominals and a discussion of the modeling of their diachronic development from the perspective of Functional Discourse Grammar (FDG). It starts with an outline of the basic notions and principles of the theory of FDG (Hengeveld & Mackenzie Reference Hengeveld and Lachlan Mackenzie2008). A key feature of FDG is its top-down approach, starting with the speaker's intention and ending up in phonological articulation. A second basic property is its function-to-form orientation, meaning that it only accounts for linguistic phenomena that are systematically encoded in the grammar of a language. These two premises are reflected in the model's architecture: the speaker's communicative intention on the prelinguistic level in the Conceptual Component feeds into the Grammatical Component, where a mental representation is translated into a pragmatic representation at the Interpersonal Level (IL) and a semantic representation at the Representational Level (RL). These configurations are then encoded at the Morphosyntactic Level (ML) and, finally, at the Phonological Level (PL). FDG's distinction of different levels of analysis and the interfaces between them is a crucial tool that allows us to account for the development of the various non-prototypical of-binominals through the changing relationships between elements at the different levels. Another relevant theoretical principle of FDG is the distinction between lexical and grammatical elements, although FDG acknowledges that the distinction is fuzzy.

A final important principle of the theory of FDG is transparency, that is, the one-to-one correspondence between units of meaning and units of form. In the prototypical N+PP construction (e.g. the whale of the North), the speaker picks up an entity for reference and specifies it with a location. During formulation, the speaker's intention is structured as two Referential Subacts at the IL, corresponding to an Individual and a Location at the RL. The Individual (‘whale’) represents the referent of the whole expression and the preposition of is a lexical item indicating origin or source. The N+PP construction exhibits a high degree of transparency, with a correlation between Referents at the IL with an Individual and a Location at the RL. There is also a transparent relationship with the units at the ML, as the two Referential Subacts are realized as a Noun Phrase and an Adpositional Phrase.

The degree of transparency between the IL and the RL decreases as we move along the path of grammaticalization from the N+PP construction to the head-classifier (e.g. a huge whale of rocks). In this construction, the speaker's intention changes, as he/she now wishes to classify the referent of the first noun, so that the second Referential Subact at the IL is lost, being analyzed as a property by means of an Ascriptive Subact. However, this Ascriptive Subact corresponds to a class of individuals on the RL. This mismatch between the IL and the RL has formal repercussions at the ML, where the second Ascriptive Subact appears as a bare noun. Another difference from the previous construction is that the preposition of does not appear at the RL, as it has no semantic content, functioning as a grammatical linking device. Additionally, the mismatch between the IL and the RL has an effect on premodification patterns. Thus, the second noun can no longer be modified by operators or modifiers of the Referential Subact.

In the EBNP construction (e.g. a whale of a wolf), transparency between the IL and the RL has been re-established. Now the first noun ascribes a property to the referent of the second and has lost its referential value, expressing an Ascriptive Subact at the IL that is aligned with a Property at the RL. It has become dependent on the second noun, which expresses a Referential Subact at the IL and corresponds to an Individual at the RL. However, the EBNP creates a mismatch between the formulation levels and the ML, as a property that fills a modifier slot at the RL (‘whale’) is realized as a nominal word within a noun phrase. The second noun is now the head and selects the first determiners, and premodification shifts into the first noun premodification zones, even when selected by the second. The preposition of is still present at the ML, as the two nouns do not form a compound, as well as the indefinite second determiner.

The mismatch between formulation levels (IL and RL) and the ML in the EBNP has been resolved in the EM construction (e.g. a whale of a time). In this case, the modifier phrase has lost its semantic content and functions as the speaker's subjective evaluation of the head. From the perspective of FDG, the first noun is analyzed as a secondary lexical element with a reinforcing function of an Ascriptive Subact, whose interpretation is context dependent. This first noun is absent from the RL, as it has no semantic content and is triggered directly at the ML. Finally, the loss of meaning and function of the preposition and the second determiner allows for phonological/orthographic reduction (e.g. a whaleuva job), which means that the EM is no longer an of-binominal but a simple noun phrase.

As is the case with the EM, the BI (e.g. a whale of an ambitious goal) is a simple noun phrase, where the first noun is a lexical operator that reinforces an Ascriptive Subact at the IL. Because this noun has no semantic content, it does not appear on the RL and is directly encoded on the ML as a [N of (a)] chunk, which can be orthographically reduced (e.g. a helluva wonderful country), re-establishing transparency at all three levels.

FDG architecture can plausibly capture the distinctions and changes between the different of-binominals. However, the appearance of this second indefinite determiner on the ML and its function in the EBNP cannot be adequately explained. In this regard, ten Wolde suggests using the Combination of Partially Instantiated Frames (ComPIF), which, unlike constructions, are restricted to multi-word expressions which are located between traditional lexemes and fully productive (abstract) frames. The ComPIF is, in ten Wolde's view, an adequate tool to explain both the regular and irregular features and avoids the introduction of constructions within the theory, which would imply denying the motivated link between meaning and form advocated by FDG.

Chapter 10 is a brief final reflection of the two main contributions of the book, that is, the differentiation of different types of binominals that appear to be historically related, and their explanation using the notions and mechanisms of two different albeit compatible theories. Ten Wolde lays out the commonalities and the differences between the two theories of CG and FDG, and highlights the strengths and the weak points of each theory. Both theories are located in the cognitive functional space, perceiving language as a tool used in social interaction and assuming that linguistic form is predominantly shaped by pragmatics and semantics. They both believe that language is based on cognitive principles, they are both usage-based and acknowledge that the lexical–grammatical distinction is gradual, although FDG still maintains this distinction. Furthermore, both theories have very similar classification methods, although FDG is more radical in considering new categories only those semantic and pragmatic changes (or differences) that systematically result in morphosyntactic changes. In this connection, FDG follows a top-down approach in that it only accounts for semantic and pragmatic changes triggering formal differences, ascribing a motivated link between meaning and form, unlike CG. Another important difference is that FDG provides grammar-internal explanations for phenomena, so that it allows us to explain the distinctions between the different of-binominals using the linguistic tools that are already available in the theory. However, FDG fails to fully account for some aspects, like the reappearance of the second determiner in the EBNP, which can be done from a network perspective.

This network view is precisely the most important contribution of CG, particularly when looking at the diachronic development of linguistic units and at the competition between two forms. Focusing considerably more on the cognitive processes, CG associates linguistic phenomena with more entrenched patterns in the language system, being able to explain the co-evolution of constructions and suggesting that form, and not only pragmatics or semantics, can also play a role in morphosyntactic changes. Additionally, the reappearance of the indefinite determiner in the EBNP can be plausibly explained by the persistence of the inheritance link to the NP-of-PP construction. In contrast to the motivated link between meaning and form postulated in FDG, CG assumes that this link is arbitrary, so that constructions are classified with dictionary-like entries. Ten Wolde finishes this chapter by advocating the integration of the network analysis into an FDG analysis for future research. Furthermore, the author suggests several subjects open to research and discussion.

In conclusion, no previous study of the EBNP to date has offered the depth of detailed description, accurate analysis and comprehensive approach – both synchronic and diachronic, quantitative and qualitative – as the present book. Drawing on authentic data, it distinguishes between constructions often overlooked and sheds light on the constructional network they belong to. Simultaneously, it draws conclusions on their diachronic development and provides plausible analytical proposals from compatible theoretical models.