An intrinsic relation between S↴S and ↴S is a relation brought about by the relationality of either S↴S or ↴S. That is, there is no relator between them that would not be an intrinsic component of either.¹ For an intrinsic interpropositional relation, this means that one proposition functions as an argument of some relational element – typically, the predicate – of the other proposition. The proposition functioning as argument then is the subordinate proposition, while the other one is the superordinate proposition. In this section, S↴S represents the superordinate, ↴S the subordinate proposition.

In a subset of the relations treated here, ↴S does bear a relation to S↴S as a whole. This is the case of the enunciative relation as long as it is not systematized as the relation of a complement clause. Moreover, it is arguably the case of the relation of the subordinate bare proposition of a phase predicate or modal predicate. In all the other cases, ↴S does not bear a relation to S↴S, but to a proper part of S↴S. Since the difference does not matter in some cases, it is ignored in the use of the term ‘interpropositional relation’.

↴S in the intrinsic interpropositional relation is generally at some level below the speech act. However, the enunciative relation freely links speech acts. Apart from this special case, this section is subdivided as follows:

1. ↴S is a bare proposition as opposed to a typed proposition.
2. For typed subordinate propositions ↴S, a distinction applies according to their basic illocution (by intersection with the functional domain of modality and illocutionary force).
3. in all subsections, a subdivision by semantic class of predicate of S↴S applies (by intersection with the functional domains of predication and of participation) (cf. Noonan 1985).

Semantic parameters that distinguish subtypes of junction beneath this subdivision derive from those discussed in the section on interdependence of the two propositions. Clausal complements are integrated to different degrees into the main clause (see the section on fusion). For agentive main verbs, this corresponds iconically to the binding force of the higher predicate and to the integration of the two situations (Givón's [1980, 1994, §3.1] binding hierarchy).

The description of a given language will show the range of constructions possible with each predicate class. The predicates taking propositional arguments may take non-propositional, i.e. concrete, entities instead. This is entirely a question of the language-specific valency and selection restrictions of the words constituting those predicates. To give just one example: the same verb coding volition with respect to a proposition, as discussed suo loco and illustrated by a, may be used to code the desire of an object, as in #b.

This is why the valency of the superordinate verb becomes topical only at the bottom of an onomasiological description of junction.

¹ To avoid misunderstandings: An interpropositional relator is typically coded by a conjunction, which is generally part of one of the two clauses. The above condition refers to the propositions, not to the clauses.