One of the challenging features in the classification of of-binominals is distinguishing between the syntactically very similar forms. Therefore, Chapter 2 presents a variety of diagnostic tests in order to tease apart the constructions in the evaluative binominal noun phrase family. This chapter presents a comprehensive overview and discussion of the tests used in previous studies resulting in the following classification criteria: head status (Noun 1 or Noun 2), constituency tests, constraints on the selection of the two nouns, determiner selection restrictions, and the status of the preposition of. These tests are then used to classify all six of the of-binominals addressed in this study.

The classification criteria in Chapter 2 are now applied to the final three evaluative of-binominals. The evaluative binominal noun phrase (an egg of a head) is an of-binominal construction in which the first noun ascribes a property to the second. The second noun is head, and the construction exhibits a number of noncanonical syntactic features, e.g. the first determiner has scope over the whole construction, a restricted second determiner, and irregular premodification patterns. The evaluative modifier (a whale of a time) is a new of-binominal that I propose. In the evaluative modifier, the first noun has completely decategorizated and functions as a part of the [N1 of a] chunk that denotes speaker evaluation of the referent denoted by the second noun. Furthermore, I argue that this construction needs to be distinguished from the EBNP. Finally, in the binominal intensifier (a beast of a good read), [N1 of a] functions as an intensifier or booster, modifying the gradable adjectives that follow. This study demonstrates that this final construction is much more prolific than previous research has shown.

The three case studies presented in this chapter demonstrate that the six of-binominals introduced in Chapters 3 and 4 form a grammaticalization path, starting at the N+PP and ending, in most cases, at the BI (cake is the exception). The chapter begins with a discussion about the differences between grammaticalization and lexicalization, since both processes are plausible in this case. Then, looking at the first nouns beast, cake and hell (an animate, inanimate, and abstract first noun respectively) and using a range of historical corpora, this chapter presents a qualitative diachronic analysis that looks at first attestations of and discusses the use of these first nouns in the six of-binominal constructions presented in Chapters 3 and 4. Ultimately this chapter substantiates the claim that first nouns progress from the N+PP to the head-classifier, in some cases pseudo-partitive, then the EBNP, the EM, and the BI. Furthermore, it argues that the process demonstrated is indeed grammaticalization rather than lexicalization.

The chapter begins with an introduction to the FDG theory and language model. The discussion includes some of the issues in the classification of premodifiers and compares the distinctions made in FDG with the construction-based model. The second section presents an FDG model of each of-binominal, illustrating how the model captures the differences between these constructions. By doing so, current issues in recent FDG research are addressed, such as the role of CxG like constructions in FDG, interfaces and mismatches between levels, and the distinction between operators and modifiers. One important finding is that FDG predicts the changes of premodification patterns found in Chapter 7. The historical analysis in the FDG frameworks shows that we are looking at the reduction of the internal components of the N-of-N template to a simple NP template, with a reanalyzed chunk [N of (a)], functioning first as a modifier and then as a lexical operator. Therefore, although we are structurally looking at an of-binominal, this term is actually a misnomer

The chapter begins with a brief overview of research on the of-binominal, in general, before focusing on the constructions addressed in this study. It presents the constructions discussed in this work: prototypical N+PP, head-classifier, pseudo-partitive, evaluative binominal noun phrase, evaluative modifier, and binominal intensifier, and the research gap that the monograph fills. Furthermore, it briefly introduces the two theories used: Construction Grammar (CxG) and Functional Discourse Grammar (FDG) and presents the rationale behind examining the results from two theoretical paradigms. This chapter, furthermore, explains the choice of corpora used in the research (primarily the Corpus of Contemporary American English (COCA) and the Corpus of Historical American English (COHA) and details the construction of the dataset: those constructions included and excluded from the data set.

Chapter 8 examines the findings from both Part I and Part II of this study from a CxG perspective. It begins with an introduction to CxG theory, constructional network models, and the most recent work on horizontal relationships in the network. I argue that the most interesting aspect of the CxG account is the network perspective on language change; this allows for speculation on how the language system might have influenced syntactic and semantic changes found. Furthermore, it is shown that both the allostruction model and multiple inheritance links are needed in order to explain the changes found in this grammaticalization path. In particular, the CxG analysis posits a link to the simple NP frame starting with the head-classifier, which would account for the internal changes that took place in the EBNP, evaluative modifier, and binominal intensifier. The persisting inheritance link between these constructions and the N+of-PP schema would offer a plausible explanation for the reappearance of the indefinite determiner in the EBNP. Competition between these two schemata, N+of-PP and simple NP, would explain the formal and functional changes found in the grammaticalization path.

Chapter 6 presents a more recent quantitative diachronic study looking for historical evidence for the grammaticalization path presented in Chapter 5. Using data from COHA and COCA, the study examines the development of the first nouns nub, breeze, husk, snake, bitch, and whale. The chapter begins with a discussion of bridging contexts as evidence for the constructional changes. The methodology section provides an explanation for both the choice of first nouns and the choice of corpora. The analysis is broken down into different paths: those first nouns that developed a pseudo-partitive use before the EBNP (nub, breeze, husk, and snake), the pseudo-partitive path, and those first nouns not used in the pseudo-partitive and which developed an evaluative meaning after the EBNP, the evaluative path (bitch and whale). The evidence supports the grammaticalization path proposed in Chapter 5 and indicates that the pseudo-partitive plays only a subsidiary role in this grammaticalization process; it does not appear to be directly linked to the EBNP. The more plausible historical link is between the head-classifier and the EBNP.

This study has proposed that [N of a] unit in the evaluative modifier and binominal intensifier, and to some degree the evaluative binominal noun phase, shares formal and functional features with premodifiers. If this is true, this unit should have integrated into the preexisting premodification patterns in front of N2, and evidence of this change can be shown in the distribution of the premodifiers in the six of-binominals. This is tested using a dataset compiled from COCA consisting of 3,433 premodifier of-binominal tokens and Ghesquière’s (2014) zone-based premodifier classification. A Hierarchical Configural Frequency Analysis was conducted to explore the premodification patterns of each construction. The findings show that all six of-binominals demonstrate different premodification patterns and corroborates the findings that each are separate constructions. Furthermore, the evidence substantiates the claim that [N of a] forms a separate unit and is situated in the appropriate place in front of N2. It also demonstrates that these two constructions follow the grammaticalization path of more canonical premodifiers.

The final chapter briefly summarizes the key findings in Parts I and II and compares and discusses the strengths and weaknesses of the models discussed in Chapters 8 and 9. In particular, it is argued that one of the fundamental differences between these two approaches is the information they seek to model. FDG offers defined primitives and combinatorial constraints that function as a basis of analysis and constrain possible outcomes. In the context of this project, this means that FDG allows us to capture the distinction between the six of-binominal categories discussed by using the language-specific tools that already exist in the model. However, the FDG account lacks a network view of these phenomena. The CxG analysis offers a network perspective on the changes in constructions and links these constructions to more entrenched patterns in the language systems. This means that it can capture the co-evolution of constructions in the language system. Finally, this chapter discusses some remaining open questions and suggests potential avenues of future research.

In this chapter, the classification criteria described in Chapter 2 are applied to the first three of-binominals in order to detail the formal and functional properties of each construction. The first construction is the prototypical N+PP (a whale of the north). This category represents an amalgamation of of-binominals all of which share syntactic features that could be regarded as characterizing canonical of-binominals in English, e.g. the first noun is the head, and the second noun is part of an NP embedded in a prepositional phrase functioning as a post-modifier. The second is the little researched head-classifier (a beast of prey). In this construction, the prepositional phrase qualifies the first noun; the second noun does not refer to a discourse referent and instead classifies noun1. This study distinguishes between two types of head-classifiers: taxonomic and intrinsic head-classifiers. The last construction is the pseudo-partitive (a cup of coffee). In this of-binominal, the second noun is head, and the first noun is a relational noun that measures/quantifies or indicates the shape of the second noun.