The goal of this chapter is to develop a theory of Formal Features that will capture and unify many of the generalisations we have arrived at in previous chapters and, more specifically, to develop a theory of Case, agreement and movement, showing how these three notions are intertwined. The core notion is the Agree relation.

In this chapter, we look at empty pronouns: cases of ‘silent deep anaphora’. More specifically, our focus is on null subjects, i.e. silent pronominal subjects, in a range of languages. We identify three main types of null-subject languages and attempt to explain the typology. This leads us to the investigation of the parametric variation across languages.

The chapter provides an introduction to Systemic Functional Grammar, in the context of Systemic Functional Linguistics. It introduces the central concept of choice between language features and the use of networks to model that choice. One such choice lies between process types and this is explained as important to the concept of semantic field used in the book. Systemic Functional Grammar describes the relationship between lexis and grammar as a continuum; one challenge is to specify how the grammar and specific lexical items relate to one another. Three responses to this challenge are discussed: Hasan’s use of system networks to distinguish between the features of near-synonyms; Matthiessen’s use of Levin’s verb classes to add detail to the notion of process type, and the approach taken by the Cardiff grammar, specifically by Fawcett, Neale, and Chrispin.

The chapter summarises the study described in the book. It discusses the contribution of the study to Construction Grammar and the applications of this grammar to discourse analysis and to language teaching. The contribution of the study to Systemic Functional Grammar is then discussed, with a comparison between this study and proposals by Halliday and Matthiessen. The chapter concludes with a discussion of how the study might be extended in future work.

The chapter introduces the idea of using networks inspired by Systemic Functional Grammar to model the relationship between constructions that express a semantic field. The example used in the chapter is Causation. From the work described in chapter 3, 105 verb argument constructions are identified as expressing Causation. A preliminary division is made between causing an action, causing a state, and causing a change in thought or emotion. Finer distinctions within each are then identified, with these distinctions being explained and modelled in taxonomies or Meaning Networks. Finally, the features that contribute to those networks are identified and expressed as independent choice networks, referred to as Systemic Networks.

The chapter explains the process of building Meaning Networks and Systemic Networks, as described in chapter 6, for two semantic fields: Cognition and Communication. The identification of these fields is inspired by the Systemic Function Grammar processes: mental and verbal. The Cognition field is divided into Emotion (53 constructions), Perception (9 constructions) and Thought (92 constructions). Following an overview, the Communication field is divided into communication about a future action (Communication: Action) (21 constructions) and communication about information (Communication: Information) (82 constructions). For each semantic field, the constructions are described as they relate to one another. Their significant features are identified and expressed in Systemic Networks. The distinctions or choices between the constructions are modelled in taxonomies or Meaning Networks.

The chapter explains the process of annotating constructions for semantic roles. Constructions expressing nine semantic fields have been annotated, with each noun phrase, adjective phrase, or clause in the construction given one of a finite set of labels. The process is compared with approaches to role identification taken in Corpus Pattern Analysis (Hanks), Systemic Functional Grammar (Halliday and Matthiessen), Local Grammar (Barnbrook, Su) and FrameNet (Fillmore). The role names used in this book, with their definitions, are listed for each of the nine semantic fields used as examples.

The chapter describes the process by which 50 verb complementation patterns have been reinterpreted as verb argument constructions, leading to the identification of 800 constructions. The chapter gives examples from seven patterns: V into n; V n, V n adj; V n to n; V after n; V n that and V n with n. The constructions derived from each pattern are arranged in networks inspired by those used in Systemic Functional Grammar. The networks show the similarities and differences between constructions. They can also be used to show constructions at broader and narrower levels of generality.

The chapter introduces the key concepts in the book. It explains corpus lexicography and the concept of Pattern Grammar, Construction Grammar and its relation to verb argumentation, and Systemic Function Grammar. It illustrates how the three can be brought together, unifying disparate approaches to the description of English. The chapter includes notes on the websites, corpora, and conventions used in the book.

The chapter explains the process of building Meaning Networks and Systemic Networks, as described in chapter 6, for four semantic fields inspired by the concept of material process and a further two semantic fields inspired by the concept of relational process. The fields are: Change, Creation, Location_change, Possession_transfer, Equivalence, Logical_relation. For each semantic field, the constructions are described as they relate to one another. Their significant features are identified and expressed in Systemic Networks. The distinctions or choices between the constructions are modelled in taxonomies or Meaning Networks.

The chapter provides an introduction to the concepts of Pattern Grammar and Construction Grammar, with a discussion of their similarities and differences. Pattern Grammar refers to a notation system devised to record, in a transparent and flexible way, the behaviour of individual words, as part of corpus lexicography. It has been found that words sharing patterns can be grouped according to shared aspects of meaning. In Construction Grammar, constructions are proposed as pairings of form and meaning. Most constructions allow for variability in the words used in them, with meaning belonging to the construction rather than to the word. Thus, both approaches link form and meaning. Many observed phraseologies can be interpreted both as examples of grammar patterns and as instances of constructions. It is therefore reasonable to propose that the extensive corpus research underpinning Pattern Grammar can be used to inform an inventory of constructions.