Relative clause constructions express an event that functions as a modifier of a referent. As such, relative clause constructions share a participant with the matrix clause -- namely, the referent that they modify. Like other complex sentence constructions, relative clause constructions may be balanced or deranked. The primary differences in strategy involve the expression of the shared participant. The most common strategy is the externally headed strategy: the referent is expressed as a common noun phrase in the matrix clause, and in reduced or zero form in the relative clause. A small minority of languages use strategies that appear to form a continuum from internally headed to correlative to adjoined constructions. Events that function to modify a referent that is a very peripheral participant in the relative clause events form noun-modifying clause constructions; these constructions sometimes use a relative clause strategy. Relative clause construction strategies also systematically vary with respect to the semantic role of the referents in the relative clause event, which are ranked by the Accessibility Hierarchy. Relative clauses may have an anaphoric or indefinite head.

A predication prototypically predicates an event. Events have multiple participants in their semantic frame. Some participants are more central than others. The information packaging of event participants construes certain participants as core arguments and others as oblique arguments. Transitivity constructions are defined in terms of the prototypical expression of central participants as core argument phrases. ‘Subject’ and ‘object’ are defined crosslinguistically in terms of degree of topicality (salience) and force dynamics (subject acting on object). Basic argument encoding strategies are flagging, indexation, and word order. An exemplar approach to defining transitive constructions is taken, using the agentive change of state event of breaking as the exemplar event, following Haspelmath. Subject generally precedes verb and object in word order. Variation in alignment is based on the system of transitive and intransitive constructions, in terms of which core argument of the transitive construction the intransitive argument aligns with, including the rare case where the core arguments of intransitive constructions are split between transitive subject and object.

Speech act constructions bear a close functional relationship to modality and polarity, and also to the information packaging of clauses (Chapters 10–11). Declaratives are associated with the modal category of polarity: declaratives assert or deny the truth of a proposition. Interrogatives (questions, and also responses) are associated with identificational packaging: the information asked about is the focus. They are also associated with epistemic modality: they involve degrees of (un)certainty about an event. Imperative--hortative speech acts are associated with deontic modality: both express a future event that is being at least considered by an agent. Exclamations are associated with the mirative (expression of surprise), which in turn is associated with thetic information packaging. These functional relations between speech act, modality, and clausal information packaging are manifested in the sharing of morphosyntactic strategies between the related categories.

In computer science, a graph means a network: a collectionof things (people, web pages, subway stations, animal species, . . .) wheresome pairs of those things are joined by some kind of pairwise relationship(spent more than 15 minutes inside an enclosed space with, has a [hyper]linkto, is the stop before/after on some subway line, is a predator of, . . .).It’s possible to make graphs sound hopelessly abstract and utterlyuninteresting—a graph is a pair〈V, E〉, where V is a nonemptycollection of entities called nodes and E is acollection of edges that join pairs ofnodes—but graphs are fascinating whenever the entities andthe relationship represented by the edges are themselves interesting!Here are just a few examples.

Complement clause constructions express an event that functions as a participant in another event -- expressed as the complement and the complement-taking predicate (CTP), respectively. Complement clause constructions often differ depending on the type of CTP, and sometimes by the factivity (epistemic stance) of the complement event. Semantic types of CTPs form a hierarchy, the Binding Hierarchy, in terms of whether their complement will be expressed by a balanced or deranked dependent clause construction. Balanced complement clause constructions may originate in independent clauses, particularly direct speech complements, and spread down the Binding Hierarchy; some deranked complement clauses originate in purpose adverbial clause constructions. Complement clauses may share participants with the CTP event; this is inherent to CTP meaning at the lower end of the Binding Hierarchy, which includes TAMP forms. The argument structure constructions associated with complement clause constructions may reflect sharing of participants through partially or fully merged argument structure strategies, or via logophoric constructions.

Reference can be done by words defined by type (common nouns), token or individual (proper nouns), or contextually (pronouns). Reference in these three ways is almost always to individuals. The animacy of common noun categories is often relevant for grammatical behavior. Personal pronouns and demonstrative pronouns are defined by properties of the speech act context. Contextual expressions may stand alone for reference or function as modifiers of nouns, i.e. attributives or articles. Articles are defined by two subtle contextual properties, referent status and identifiability. Referent status involves accessibility in discourse or shared knowledge, and, for non-accessible referents, whether they are real or not. Identifiability pertains to whether the referent’s identity is known, or is only identifiable by type. Distribution of pronoun/article uses can be represented as semantic maps on a crosslinguistic conceptual space of functions. Tracking of a referent in discourse is grammatically encoded as often as referent accessibility or identifiability. Finally, reference to a type (generic) reference is possible; strategies are typically recruited from reference to a token.